1 Deliverable D 9.2 Safety Standards for Road Design and Redesign SAFESTAR FINAL REPORT Public Contract No. RO-96-SC.203 Project: Safety Standards for Road Design and Redesign Coordinator: SWOV Institute for Road Safety Research, Leidschendam NL Partners: TNO Human Factors Research Institute; Soesterberg, NL RD Road Directorate; Copenhagen, DK VTI Swedish NationalRoad and Transport Research Institute; Linköping, S VTT Technical Research Centre of Finland; Espoo, FIN LNEC Laboratório Nacional de Engenharia Civil; Lisbon, P NTUA National Technical University of Athens; Athens, GR CETE Centre d'Etudes Techniques de l'Equipement Normandie Centre; Grand-Quévilly, F CDV Transport Research Centre; Brno, CZ Date: November 2002 PROJECT FUNDED BY THE EUROPEAN COMMISSION UNDER THE TRANPORT RTD PROGRAMME OF THE FOURTH FRAMEWORK PROGRAMME
Transcript
1
Deliverable D 9.2Safety Standards for Road Design and Redesign
SAFESTAR
FINAL REPORT
Public
Contract No. RO-96-SC.203
Project: Safety Standards for Road Design and RedesignCoordinator: SWOV Institute for Road Safety Research, Leidschendam NL
Partners: TNO Human Factors Research Institute; Soesterberg, NLRD Road Directorate; Copenhagen, DKVTI Swedish NationalRoad and Transport Research Institute;
Linköping, SVTT Technical Research Centre of Finland; Espoo, FINLNEC Laboratório Nacional de Engenharia Civil; Lisbon, PNTUA National Technical University of Athens; Athens, GRCETE Centre d'Etudes Techniques de l'Equipement Normandie
Centre; Grand-Quévilly, FCDV Transport Research Centre; Brno, CZ
Date: November 2002
PROJECT FUNDED BY THEEUROPEAN COMMISSION UNDERTHE TRANPORT RTD PROGRAMMEOF THE FOURTH FRAMEWORKPROGRAMME
SAFESTAR was a research study focusing on traffic safety for what is known as the ‘Trans-European Roadway Network’ (TERN) that links the major European centres. The knowledgeneeded for being able to carry out an effective safety policy at the European level is insufficientin regard to various safety aspects of road infrastructure. SAFESTAR was established to fill inthese gaps of knowledge, with special notice being given to the following seven topics:emergency lanes and shoulders along motorways, tunnels located on motorways, express roads,cross-sections of rural roads, curves in rural roads, major junctions on roads in urban areas, andassessing the safety of road infrastructure during the planning and design stages (the performingof safety audits).Within the scope of SAFESTAR, the investigation into these areas pays much attention to thedifferences between design standards as they now exist or are being developed in the individualcountries within Europe. Of particular interest are the differences between the countriesparticipating in SAFESTAR: the Netherlands, Denmark, Sweden, Finland, France, Portugal,Greece and the Czech Republic.
- SAFESTAR is providing an overview of the nature and the degree of danger in theemergency lanes and shoulders of European motorways. Next, an inventory and analysis arebeing made of the various measures currently taken to correct these problems. Finally, thestudy goes deeper into the criteria for applying various kinds of safety provisions such as safetybarriers.- SAFESTAR will select a tunnel design deemed responsible by expert opinion. Next, thistunnel design will be tested in a driving simulator.- SAFESTAR is charting the dangers of express roads and in doing so, analyses thereasons for choosing this type of road. Attention is also given to the future development ofexpress roads: will the problem disappear due to the construction of additional (semi-)motorways, or on the contrary, will the situation worsen due to the increasing amount oftraffic? Finally, recommendations are being provided for road design and for the circumstancesunder which these roads should be constructed.- The cross-section of a road determines to a great extent which traffic situations can occuron such a road and also which accidents can occur. SAFESTAR evaluates the difference incross-sections based on the characteristics of the accidents. A few promising correctivemeasures are being selected, and the effects of these are being studied with test subjects drivingvehicles equipped with test devices.- Another task of SAFESTAR is to increase knowledge in the effect that the design ofcurves has on the safety of rural roads. The evaluation is based on two methods: by usingcalculation models for the speed profile and accident frequency, and by investigating drivingbehaviour immediately before and in curves when using different kinds of marking and signing.- For designing urban junctions, it would be desirable to have a calculation model whichwould predict accident levels that will occur once the junction is built or put back into use aftermodifications. These calculation models already exist, but they still require a great deal ofimprovement. SAFESTAR focuses on obtaining an improved calculation model.- The aim of safety audits is to assess the pre-construction safety level of roadinfrastructure design. A design can be assessed in various phases of the design process. Teams
4
of experts not directly involved in the project carry out the audits. SAFESTAR will evaluateexisting audits procedures and will test a number of different procedures in different countries.
The findings produced by SAFESTAR will be compiled to create a coherent list accompaniedby recommendations concerning the safety aspects of road design.
5
EXECUTIVE SUMMARY
The level of road safety is, to a large extent, determined by the features and layout of the roadtransport system infrastructure. If the human errors which result in accidents are to be held incheck then proper road design is crucial. it has been estimated that improvements in theengineering of roads has been one of the main factors behind the reduction in casualties on theroads of EU countries in recent years.
To achieve their full effect safety principles in road design have to be applied in a systematicand consistent manner. Progress towards the optimal adaption of road design to theseprinciples is expected to produce a considerable reduction in the number of accidents andaccident rates compared to the existing situation in Europe.
Standards play a vital role in road design. Not all countries have a full range of design standardsapplied to their road networks and this situation contributes to the size of the road safetyproblem on the continent as a whole. Continued improvement of road design standards on theTrans European Road Network (TERN) is required and this will help to install good practiseon all types of road throughout Europe.
Proposals and agreed technical standards, however, cannot be expected to flow simply from asafety perspective. The overall objective of the SAFESTAR programme has been theformulation of safety arguments for selecting particular design elements or dimensions forinclusion in the improvement and augmentation of design standards.
The safety arguments produced in the course of this study do not lend themselves to summaryand simplification. They are laid out, section by section, in the main body of this report. Thestandards derived from them are listed, for convenience, in Section II. However, a number ofthemes, flowing from the work, are discussed briefly below on a topic by topic basis.
The list of standards in Section II cannot be considered complete because the researchreviewed and carried out for the project could not fill all the gaps in our present knowledge. Asecond list indicating where more data and research are required can be found in Section III.
Hard shoulders (Emergency lanes) on motorwaysCurrently, there is a safety problem because of the design of hard shoulders and the presence ofobstructions on or adjacent to them. SAFESTAR has provided an overview of the nature anddegree of danger on hard shoulders. Current measures taken to combat the problem have beenreviewed and analysed. Criteria have been generated for the application of safety barriers and other measures.
Tunnels on motorwaysFrequently the dimensions and design standards of tunnels do not match those of the adjacentstretches of motorway and this has resulted in problems. SAFESTAR has selected a number oftunnel designs intended to reduce problems and has tested these designs by means of drivingsimulators.
6
Express RoadsThe current infrequent application of this type of road on TERN is expected to increasesignificantly in the near future. Because it caters for long distance and local traffic it is knownto be relatively unsafe. SAFESTAR has catalogued the dangers and analysed the reasons forthe choice of this type of road, forecasting developments and recommending standards ofdesign, and a basis for the choice.
Rural RoadsCurrently about one third of the length of TERN is comprised of this type of road which isknown to generate the vast majority of injury accidents outside built-up areas. SAFESTAR hascarried out evaluations and looked at research on the different effects of corrective measuresand a variety of marking and signing with reference to cross-sections and to the design ofcurves.SAFESTAR has developed improved calculation models which make it possible to assess curvedesigns for safety aspects before the roads are constructed.
Major Urban JunctionsA large number of designs and design methods for dealing with these junctions is alreadyavailable. There is a need for improvements to the process of choice of design with reference tothe safety of all road users including cyclists and pedestrians. SAFESTAR has focused onobtaining improved calculation models.
Road Safety AuditsThe principles and practise of Road safety Audits (RSA) are seen as an excellent tool forimproving safety through the careful monitoring of design by independent experts. SomeEuropean countries already have procedures for carrying out RSA. As part of SAFESTAR,these procedures have been described and compared. RSA also appears to offer an opportunityto promote consistency in design standards. (see Section I)
7
I. RECOMMENDATIONS
I.1 Recommendations for the European Commission (DG VII/Transport)
� The European Commission should sponsor a pilot scheme of Road Safety Audits (RSA)on construction and upgrading on TERN roads. It should also act as broker for thedevelopment of a widely acceptable protocol for RSA which should include standardsand procedures. The protocol should also state the requirements for, and theresponsibilities of the audit team.
� The communication and dissemination of knowledge about the safety aspects of roaddesign are a pre-requisite for obtaining a (sustainably) safe road network. TheCommission should play a central role in the creation of the tools and the channels toachieve this on a pan-European basis.
� The Commission should organise the preparation of a code of good practise for thedesign of roads in the Trans European Road Network.
� Although there has been some progress in the creation of pan-European research anddata-logging it is important that future design and regulation does not depend on thecompilation of the findings of disparate research projects. The Commission shouldactively stimulate co-operation in research to a level where it will support policy anddesign standards across Europe.
I.2 Recommendations for National governments (Departments of Transport)
� National goals for the improvement of road safety should be set as an essential element ofroad safety policy
� The appropriate resources to facilitate the achievement of these goals should then beassigned to Road Authorities, those responsible for design guidelines and manuals andResearch Institutes working in this field. This should include budgets, laws, regulations,organisation and the provision of tools for the design of roads.
� The development of Road Safety Audit procedures and methods should be stimulated andthe results applied in practise.
8
I.3 Recommendations for Road authorities and those responsible for manuals andguidelines on road design
� Three basic design principles should be adhered to:� The road system should be classified into types with different design standards� Design should seek to prevent situations in which large differences in speed, mass and
direction are present simultaneously.� Design should facilitate the predictability of behaviour in traffic situations.
� At a national level the principles and practise of Road Safety Audit should be integratedinto the design and construction process as a tool for improving road safety.
� The new safety standards and procedures derived in this research should be applied toroad schemes and incorporated in manuals and guidelines. (see Section II)
I.4 Recommendations for Research Institutes
� National road research institutes should evaluate Road Safety Audit both with referenceto results obtained in the actual application of the technique and the potential for savingsto flow from its general application in the design process.
� Efforts should be made to improve the involvement in cross-national and pan-Europeanresearch to support the creation of a full range of European policies and standards.
� When planning research programmes, reference should be made to the list of new andadditional research requirements which is included in this report (Section III).
9
II. NEW SAFETY STANDARDS SUGGESTED BY THIS RESEARCH
In this list the new standards are organised by road type for convenience. Within the main bodyof this report they are organised by topic in Sections 2.1 to 6.2. where they appear alongsidediscussion of the research from which they are derived and the reasons why they arerecommended.
II.1 Motorways
The design of motorways should incorporate the following:
• An obstacle-free zones at least 9 metres wide on each side of the carriageway. (seeSection 4.1)
• Any inclines within these obstacle-free zone should not be steeper than 1 in 5 (20%) forslopes with a total height of more than 5 metres. Where the total height is 2 metres orless the slope should not be steeper than 1 in 6 (17%). (see Section 4.1)
• The median strip should have a width of at least 20 metres except where an appropriatesafety barrier is used. (see Section 4.1)
• Where sections of hard shoulder (emergency lane) are identified as having a greater thannormal risk of accidents they should be improved by an increase in width, the applicationof a rumble strip, or an improvement in lighting. (see Section 3.1)
� Where tunnels are used, the road layout within the tunnel should not be allowed to exerttoo much influence locally on the road user’s choice of speed (for example, the suddendisappearance of the hard shoulder (emergency lane) will have a substantial influence onthe choice of speed). (see Section 5.1)
• At exits and entries situated within a tunnel the road user should always have a clear viewforward of at least 100 metres. (see Section 5.1)
II.2 Express roads
The design of express roads should incorporate the following:
• Use should be restricted to high-speed motorised traffic. (see Section 3.2)
• The frequency of access and exit should be restricted. (see Section 3.2)
• Vertical alignment over the brow of a hill (convex curve) should be such that the forwardview should never be less than the distance required to stop safely. (see Section 2.1.1)
10
• Vertical alignment through a dip or hollow (concave curve) should have a minimumradius of 3,000 m. (see Section 2.1.1)
• The lane width on both single and dual carriageways should be 3.5 m. (see Section 3.2)
• The cross-section of the carriageway should include a continuation of the paved areabeyond the edge of the traffic lanes. (see Section 3.2)
• The median strip should have a width of at least 20 metres except where an appropriatesafety barrier is used. (see Section 4.1)
• Where a safety barrier is used in a median strip a recovery zone should be used betweenthe barrier and the traffic lanes. It should be wide enough to allow the recovery ofvehicles to take place. (see Section 4.1)
• The median strip should be free from slopes and obstacles. (see Section 4.1)
• Where a single carriageway road climbs a significant incline a crawler lane (climbing lane)should be included on the uphill side of the road. (see Section 3.2)
• Cuttings and embankments alongside the carriageway should not be steeper than 1 in 5(20%). (see Section 4.1)
II.3 Single carriageway rural roads
The design of single carriageway rural roads should incorporate the following
• A lane width of 3.5 metres. (see Section 3.3)
• Shoulders on each side of the traffic lanes to a width of 1.3 to 1.5 metres, giving a totalcarriageway width of approximately 10 metres. (see Section 3.3)
• Cuttings and embankments alongside the carriageway which are not steeper than 1 in 5(20%). (see Section 3.3)
• Obstacle-free zones extending for at least 3 metres on each side of the carriageway. (seeSection 3.3)
• The horizontal alignment of the road should remain consistent (as defined inSection 2.1.2).
• Road marking and signing within curves based on the strategy tested in SAFESTAR. (seeSection 2.2.1)
11
• Poor perception of a curve during both approach and negotiation should be prevented byimproved marking and signing and by cutting back vegetation which might obscure theview. (see Section 2.2.1)
• Amelioration of the speed at which a curve is entered by various devices. (seeSection 2.2.1)
• Avoidance or the reduction of consequences where a vehicle leaves the road by the use of hard shoulders, safety barriers, and high friction surfacing. (see Section 2.2.1)
• The reduction of head-on collisions by the use of ghost islands and hard shoulders. (seeSection 2.2.1)
II.4 Major urban junctions
The design of major urban junctions should incorporate the following:
• A clear view for an adequate distance for all road users regardless of weather conditionsor time of day. (see Section 6.2)
• A maximum speed differential between road users of 30 km/h. (see Section 6.2)
• A choice of junction type to maximise the effect of accident reduction. (see Table 6.10)
• The arrangement of traffic streams so as to avoid as far as possible compromising thevisibility and the prediction of the behaviour of road users. (see Section 6.2)
12
III. AREAS WHERE MORE RESEARCH OR DATA ARE REQUIRED
III.1 Motorways
• Within the SAFESTAR project itself some of the design characteristics of long tunnelshave been investigated, including the width of the hard shoulder (emergency lane),patterns on the walls and ceiling, and exits and entries within the tunnel.Other characteristics remain to be investigated:
lighting conditions, lane widths, curves, slopes, restriction of sight distances, and thevariation in traffic volumes.
• The accident risk on the hard shoulder (emergency lane) can only be established withreference to data about the number of accidents and the level of exposure to risk. There isa need for these data to be collected which is not currently being fulfilled.
• Safety barriers of various types are available. More work is needed on the comparativecost-effectiveness of the different types.
• Barrier testing should also be carried out with regard to impacts by heavy vehicles andparticularly by those with a higher than average centre of gravity.
III.2 Express roads
• If express roads continue to exist as a separate class within the road network more workwill be required on the comparative safety of motorways, express roads and other inter-urban roads.
• There is a need to identify which safety measures will improve safety on express roadswithout generating a false expectation of adherence to motorway standards.
• What long term strategy and level of standards will produce an economically acceptablelevel of cost-effectiveness for governments and authorities dealing with express roads?
III.3 Single carriageway rural roads
• The design of these roads can currently be supported by the use of speed and accidentmodels but they are in need of considerable improvement.The current data need augmentation and refinement and the variables modelled and rangeof application need to be extended and better defined.
• The effectiveness and availability of alternative methods of separating opposing trafficstreams and avoiding accidents where vehicles leave the road is another area which needmore work.
13
III.4 Major urban junctions
• Although a great deal of experience has informed the current design of major urbanjunctions there is a need for more investigative work with regard to cyclists andpedestrians especially in terms of behaviour, conflicts and accidents.
• The design of junctions can be supported by accident models. However, these models stillneed a great deal of improvement with regard to the data (number and type of junctions,number and type of accidents), the model structure (type of variables), and thespecifications for the range of application (which design elements, which stage of thedesign, which type of junction, which country).
14
1. INTRODUCTION
1.1 Design philosophy
Designing roads is a profession which needs different kinds of skills. First of all one uses thetraffic engineering techniques. Secondly one uses notions about driver behaviour, how roadusers react on the road and its environment. And last but not least, one is guided by a designphilosophy. This philosophy can be implicit, formed by intuition and experience, or it can beexplicit, developed by research and evaluations. SAFESTAR prefers to enhance an explicitdesign philosophy. In fact a combination of two philosophies: ‘sustainably safe traffic andtransport system’ and ‘relation design’. Both philosophies are described in this report.
1.1.1. The concept of a ‘sustainably safe’ traffic and transport system
The Dutch concept of a sustainably safe traffic and transport system (STTS) has far-reachingambitions. The Dutch national government has set a goal of 50% fewer road traffic fatalitiesand 40% fewer casualties by the year 2010, (taking the figures for 1985 as the baseline).Because these aims go further than merely following an existing trend, a sustainably safe trafficsystem with structurally low accident figures has to be found. An important component of sucha system is a sustainably safe design for the Dutch road network in which roads are divided intoa limited number of classes. Each class has a clear and unambiguous function for traffic andmust be easily distinguishable from the other classes.
Road safety must become an integral element of the entire Dutch road network (both urban andrural). It is inevitable that this will have major consequences for urban and regional planningand for traffic systems planning.
Starting-pointsThe STTS begins with four starting-points:- Man is the yardstick for each technical system- Adapt the traffic and transport system to man- Prevent failures- If failures do occur, then minimize the consequences
PrinciplesThe STTS gives a new meaning to the old man-vehicle-road model:- The road infrastructure should be adapted to human capabilities and shortcomings- The vehicles should simplify the driving task and should offer protection- The road users should be well-informed and, where necessary, be controlled
Design principlesThe design principles in STTS aim at connecting the three angles of the triangle FUNCTION-DESIGN-USE (the golden triangle): The transportation plan prescribes a certain function for aroad, mostly as a result from an origin-destination analysis. The traffic engineer applies thisfunction for his design. In this design he makes assumptions about the behaviour of the futureroad users (intended use). Finally, after the opening of the road, the actual use of the road will
15
show if traffic behaviour and traffic volumes agree with the original function and design. If not,the design can be altered or the use and/or behaviour can be adapted.In some cases the use which was not intended, can result in rather satisfactory situations. Inthat case the road users managed to deal with the situation. And in other cases the actual usecan be as intended but after all, not satisfactory. These discrepancies need special attention andcan teach us something more about the relations between Function-Design-Use.
Furthermore STTS wants to stimulate the interaction between the road environment and thedriving task. STTS triggers this interaction by combining unique combinations of road elementsin each road class.
Three design principles have been established:- A functionally planned road network: each link fits well into the whole system and actual
route choice is in accordance with planned route choice.- A homogeneous use of the road: road users should only be confronted with small
differences in speed and mass.- A recognizable road environment which stimulates the right expectations: predictability
of traffic situations.
Traffic and transport functionsEach traffic and transport sytem is meant to:• interconnect areas• distribute within an area• give access to ‘individual’ destinations (houses, schools, shops etc.)Of course these functions also exist in, and have to be met by, STTS.
Functional road classificationThe road classification in a sustainably safe road network differs from usual road classificationsystems. The traditional road classifications systems accept that each road can have differentfunctions at the same time. In STTS each road will get one and only one function, so thenumber of classes is the same as the number of functions:I Interconnective roads are only meant to interconnect areas.II Distributors are only meant to distribute traffic within an area.III Access roads are only meant to give access to individual destinations.A road with a certain function should be fully adapted to that function and all elements whichbelong to other functions should be removed or separated.
RequirementsThe design of the road network should meet certain requirements in order to fulfill the starting-points and principles of STTS. These requirements are of two types: functional and operational. The functional requirements can be regarded as the basic criteria for dividing the roads of thenetwork into the various classes. For each of the roads thus given a specific class, there are alsooperational requirements. These concern the most important characteristics of the cross-section, the alignment, the types of traffic (car, bike, moped, pedestrian) allowed to use theroad and their position on the cross-section. Both functional and operational requirementsshould be included in the existing guidelines for urban and rural roads.
16
Functional requirementsThe functional requirements for the road network are as follows:- largest possible areas with traffic-calming (both in rural and in urban area);- a maximal part of the journey using relatively safe roads and routes;- journeys as short as possible;- the quickest and shortest routes to coincide;- avoid the necessity to search for directions/destination;- easily recognizable road classes;- limit and make uniform the number of possible types of design;- avoid encountering oncoming traffic;- avoid encountering traffic crossing the road being used;- separate types of traffic;- reduce speed at potential points of conflict;- avoid obstacles near the carriageway.
These twelve functional requirements apply to all road classes in the entire urban and rural roadnetwork
Operational requirementsThe operational requirements have been formulated in such a way that the design specificationscan be the next step. However, the operational requirements offer many opportunities for thedesigner to vary the road design according to the local demands or according to his ownpreferences. The operational requirements must ensure that the differences between the roadclasses are bigger than the differences within the road classes. Table 1.1 shows theserequirements.
PredictibilityA small set of the fore-going operational requirements should ensure the predictability of thetraffic situations. These set comprises continuous longitudinal road elements:- marking- separation of directions- pavement, irregularity of the surface- obstacle-free zoneThe mechanism which ensures the right predictability consists of two steps: at first the roadusers must be able to recognize the road class by these four elements. Secondly, throughinformation and experience, the road user knows which possible traffic situations belong to thepresent road class.This mechanism tries to lower the workload (or mental load) of the driver. This will have apositive influence on the performance of the driving task.
WorkloadIn addition to the general effect of the operational requirements on the workload, the designercan optimize the workload by other road elements. Messer et al. (1981) have developed aprocedure to evaluate the effect of the road and road environment in the stage of the design.
17
Type of area
Rural area Urban area
Road class Interconnect-ing Road
Distributor Access Road Distributor AccessRoad
Abbreviation IR D AR D AR
Speed limit (km/h) 120/100 80 60 50 30 or less
Marking (longitudinal) fully fully but differentfrom IR
partly fully but differ-ent from IR
no
Physical separation of direc-tions (number of lanes in onedirection)
yes (1 ormore)
yes (1 or more) no (1) yes (1 or more) no (1)
Type of physical separation barrier difficult to cross n.a. difficult tocross
n.a.
Emergency facility lane shoulder or lay-by no shoulder orlay-by
no
Pavement, surface irregularity minor minor major minor major
Accesses no no yes no yes
Crossing (mid-block /between junctions)
grade sepa-rated
grade separeated at grade gradeseparated
at grade
Parking no parking lane carriageway parking lane carriageway
Public transport: stops no turnout carriageway turnout carriageway
Obstacle-free zone large medium small medium (very) small
Cyclists on the carriageway no no dependingon the local
Table 1.1 Operational requirements for five different road classes in STTS
18
In their view the workload value of a certain feature is the result of:- sight distance- expectation- unfamiliarity- workload of preceding feature- workload potential rating for the present feature
The potential rating is based on expert opinions. Krammes & Glascock (1992) showed a relationship between this workload value and theaccident risk.
1.1.2. Relation design
Lamm & Smith (1994) define ‘relation design’ as follows:“..... no more single design elements with minimum or maximum limiting valuesare put together more or less arbitrarely; rather, design element sequences areformed in which the design elements following one another are subject tospecific relations or relation ranges.”
This approach should result in a longitudinal profile which offers the car driver a consistentchain of tangents and curves. The consistency is focussed on the sight distances and designspeeds of the successive design elements.
The definition of design speed is subject to discussion. Some countries (United States,Belgium) define design speed as the maximum speed at which car drivers can use the roadsafely and comfortably, a sort of target.Other countries (Germany) accept that drivers mostly drive faster than the design speed.Therefor these countries define the design speed as the speed which is only exceeded by 15% ofthe drivers (V85). This second approach demands insight in the actual speeds which will occur.Quantitive relationships have been developed to assist this insight (e,g. Lamm & Choueiri,1987).Some countries (United Kingdom, Australia) use combinations of both definitions.
Each of these approaches demands a specific procedure to reach a satisfactory relation design.In the United States the relation design is less developed because their definition of designspeed related to each design element, results in a sequence of elements which are not very wellattached to one another. Leisch & Leisch (1977) elaborated a method to determine the speedprofile of a road. For each curve the designer determines the safe entering speed. This speed isconfronted with the speed that can be attained by the accelaration in the preceeding tangent,e.g. a curve after a long tangent will be entered with a higher speed than a curve directly afteran opposite curve. The speed profile shows the discrepancies between the safe speed and themost likely speed.The speed profile is set up for both directions, because the driving speed is depending on thesequence of curves and tangents a driver will meet.The point of the Leisch & Leisch method is that the difference between two successive entering speeds should never exceed 15 km/h. This value is based upon the experience that drivers areable to control such a speed reduction and upon an accident analysis of Glennon & Joyner(1969).
19
In Germany the Kurvigkeit or Curvature-change-rate (CCR) is used to derive the V85.One calculates the absolute sum of curvature change rates in the horizontal alignment (ingon/km2) including transition curves. The relationship between the calculated sum and V85 canbe found in the German design guidelines. This relationship was based on empirical research(Köppel & Bock, 1979). In a good design two successive elements are not allowed to showdifferences in speed exceeding 10 km/h (first condition). A second condition concerns the curveradii of two successive curves. These radii should be designed according to a relationship whichwas reported by Lippold (1996).
Both approaches, from Germany and from the United States, have their characteristicprocedures, but Lamm et al. (1986) showed that, when both methods are applied to the sameroad, the outcome will be more or less the same.
Lamm et al. (1994) promote a slightly different procedure to reach a good design consistency.Their method uses three criteria:- The difference in the V85 of two successive design elements should not be greater than a
certain value (10 km/h in a ‘good’ design and between 10 and 20 km/h in a ‘fair’ design).- The difference between V85 and the design speed should not be greater than a certain
value (10 km/h in a ‘good’ design and between 10 and 20 km/h in a ‘fair’ design) and atthe same time the radii of two successive curves should be of the same size.
- The difference between assumed and actual demanded side friction should not exceed acertain value (a positive value in a ‘good’ design and between -0,02 and 0 in a ‘fair’design).
The application of these criteria can be supported by introducing an evaluation module whichchecks each design alternative.
1.2 Road Safety Audits
This section describes tools and procedures established in different countries which conductRoad Safety Audits (RSA). These RSAs are utilized to identify potential safety problems andconcentrate on safety measures to overcome these problems. This technique is used to detectpossible safety hazards, in the various stages of a scheme, before a new road is open to traffic. The slogan ‘Prevention is better than cure’ is already well known to us, and Road SafetyAuditing can establish an association with road safety. The application of this preventivetechnique can prevent accidents or reduce the severity of accidents. Except for minimizingtrauma, and increasing the designer’s awareness of road safety, RSAs can also reduce theoverall lifetime cost of a scheme, for it is less likely that remedial rebuilding of road sectionsshould take place. Therefore this report deals with schemes subject to design and redesign ofnew roads, rather than existing roads.Strict applying of design regulations does not always lead to a safe road, for general rules don’talways fit correctly to specific situations. When applying a RSA, it improves awareness of roadsafety, and highlights safety among other aspects of road design.
Road Safety Audits’ roots
20
In use since the early 80’s, RSAs have been established as a requirement in construction andmaintenance of highway schemes in the United Kingdom . In the UK, RSA is compulsory forall trunk roads and is also issued on a voluntary basis in other road schemes. There are manyyears of experience with RSA and many countries which have developed (or which aredeveloping a RSA system) have looked carefully to the UK. Another country which has a longhistory concerning road safety is the USA. In the USA a method exists called ‘Safety Reviews’.This Federal HighWay Administration method is less formalised than the UK Department ofTransport method and also emphasises more the incorporation of guidelines and compliancewith standards rather than the use of checklists and road user behaviour. On non-FHWA roadssometimes the local road authority conducts an informal check.Over the years the evolution of auditing process has been dynamic. The idea behind safetyaudits and the scope remain the same. Since safety audits were introduced, however,experience gained from practice has been used to improve the procedures.
Objectives of safety audit as pointed out in Great BrittainThe main objective of safety audits is to ensure that highway schemes operate as safely aspossible, i.e. to minimise the number and severity of accidents occurring. This can be achievedby avoiding accident-producing elements and by providing with suitable accident-reducingelements. The purpose of safety audits is to ensure that ‘mistakes’ are not built into newschemes.Other specific aims of the Road Safety Audit are :0 to minimise accident risk on the network adjacent to new schemes0 lay emphasis on safe design practice and increase the awareness of everyone involved in
planning, design, construction and maintenance of roads. 0 to highlight the importance of taking into consideration the needs of all types of users0 to reduce the whole-life cost of the schemes, by minimising the need of future
corrections.
In order for a safety audit to be successful, certain factors should be taken into consideration.The key factors that contribute to the efficiency of the safety audit may refer to the organisationand the selection of the audit team:With respect to safety audit organisation, support and commitment of senior management isessential. Safety audits should be an integral part of an agency’s overall program. Localauthorities often use a Road Safety Plan as a framework in which the RSA is placed. By doingso, the RSA is part of the overall safety management strategy.
Undertaking the road safety audit, a procedureA RSA procedure does not differ very much between different countries. Their procedureshave many similarities with the procedure used in the UK:1. Collection of information. Such information may include: detailed plans, design
standards, traffic volumes, pedestrian counts, and accident records. At this stage, prior toany appraisal of the layout, a discussion with the design team about the objectives of thedesign, is advisable.
2. The systematic and detailed check of the design follows. However, different countries usedifferent numbers and names of stages. The UK Department of Transport Trunk roadsauditing only requires stage 1-3, the other UK stages are used by different Counties andthe IHT.
21
Number of stages used.
United Kingdom Denmark Norway Australia New Zealand
F Feasibility1 Preliminary
design2 Detailed
design3 Prior opening4 Post opening5 Maintenance /
monitoring
1 Initial design2 Preliminary
design3 Detailed
design4 Opening5 Monitoring
(Existingroad)
1 Planning2 Details3 Construction
work4 Prior opening5 Post opening
road marking /maintenance
1 Feasibility2 Draft design3 Detailed
design4 Pre opening5 Existing
road
1 Feasibility2 Project
assessment3 Final
design4 Pre opening5 Existing
roads
At Stage 2 this often involves overlaying the details from one plan on to another, as therewill be different drawings for road layout, street-lighting, safety fences, signs andmarkings. It is often the interaction of features that causes problems - for example, noone intends that lamp columns should be erected on the wrong side of safety fences. Forstages 1 and 2, following up a preliminary assessment of the design, it is essential that asite visit is carried out, so that the tie-in with existing roads can be considered, and thelocal conditions assessed. For stage 3, examination of the physical elements in site is themain task at this stage of the process. This may involve negotiating the scheme fromdifferent directions, in the dark, and under adverse weather conditions.In examining and evaluating the design, checking each element individually is one aspect.Once the audit team has predicted the type of accident problem that is likely to beassociated with an aspect of the design, a known remedy to mitigate that problem shouldbe suggested. It is important that the scheme viewed as whole and the impact of thecombination of its elements and features to the users, is taken into account.In undertaking this task, the use of checklists is strongly recommended.
3. The findings of the audit are presented in a formal audit report. A precise description ofthe possible problems identified is required, giving reasons for the anticipated conditions.For the purpose of strengthening the arguments and ensuring the objectiveness of theresults, it is required that the auditors make use of control data and the guidelines. Thefinal audit report should also include recommendations on how to solve the problem.Location plans on which identified problems are referenced, and drawings for presentingthe proposed amendments, can be used.When recommending, it should be kept in mind that the objective of the audit is theimprovement of the suggested scheme and that contradicting the designer andquestioning/changing the relevance of the design, is not desirable. Within the localsituation it is likely that the safety audit team will discuss their findings with the designteam, possibly with an informal report. This is not the case on a Road Safety Audit forthe Department of Transport where a formal audit report is required to be produced, andsent to the Department direct.
MonitoringThe use of monitoring and evaluation is a method by which road safety auditors can learn aboutcertain issues affecting the scheme, which can be applied to similar projects and areas elsewere.Feedback to the designers could lead to more awareness of the implications of their design tosafety. Nevertheless the Dept. of Transport HA 42/92 says: “It is central to the auditingprocedures that the Audit Team have no connection with the scheme design and should
22
maintain that their views are not influenced by familiarity or from natural ‘pride of authorship’".Providing safety advice by the AT to the design team conflicts with the importance ofindependence. However, many respondents to the questionaries think this kind of information ismore important than perfect independence. Training for all participants in road safety audit,including staff members, could increase the awareness and importance of safety issues. Anotherway of training the design team is providing them with checklists. By doing so, the audit teamknows which items the audit team will take notice of. Some ‘evident’ mistakes could beprevented from being incorporated in the design. This does not make the Road Safety Auditobsolete, this rather gives the AT more time to use their experience and intuition.
Some British Counties use a Road Safety Unit which is monitoring safety, and collects allaccident information (quote: ‘this keeps you rolling’, motivated) directly from the police. Formonitoring e.g. Nottinghamshire, there is also a close link between the police and the AccidentInvestigation Bureau. Roads which have been audited at the last stage, are monitored for oneyear, and after that period they are evaluated. Monitoring seems to be necessary because onedesigns for its use, but the usage changes, so the design probably should also change.
Monitoring in the consultancy branch is poor, probably the result of a lack of willingness.
ChecklistsThe purpose of the checklists is to insure that nothing is overlooked. Practitioners should notrely solely on them and are encouraged to expand them. Over the past few years checklistswere re-considered and the new checklists in the revised guidelines are meant to indicate‘principal issues’ rather than provide detailed lists of the items to be examined. Differentchecklists are provided for each safety audit stage. Checklists appear to be not very important.The usage of checklists decreases as the knowledge of Road Safety Audits increases.
Auditing existing roadsRoad safety audits refer mainly to new designs. They can however be implemented to existingroads as a complementary tool together with the analysis based on accident data. One of themain purposes when auditing existing roads is to identify if elements and features are inaccordance with the standards indicated by the hierarchy within the network.
Some elements summarised and comparedSAFESTAR has highlighted some countries using Road Safety Audits (Van der Kooi et al.,1998). In this section the differences and similarities between RSAs in those countries aresummarised. France and the USA do not have an explicit RSA system and thus are notmentioned in this overview.
Definitions used in various coutriesThe definitions used by the different countries are more or less the same. They all concentrateon road safety as one separate aspect of road use/design. The various definitions aresummarized below. It should be mentioned that the definitions regarding Denmark and Norwayare of course translations.
UK:
23
A formal procedure for assessing accident potential and safety performance in theprovision of new road schemes, and schemes for the improvement and maintenance ofexisting roads.
source: Guidelines for The Safety Audit Of Highways, IHT (1996)
Denmark:Road Safety Audit is a systematic and independent assessment of the safety aspects ofroad schemes. Its purpose is to make new and reconstructed roads as safe as possible -before construction is started and before accidents occur.
source: Manual of Road Safety Audit, RD (1997)
Norway:A systematic and independent evaluation which ensures that the products (roads, trafficsystems, traffic control) have the quality desired with respect to road safety. The auditshould not primarily control whether the planning is in agreement with traffic regulations,but should also be concerned with the auditor’s knowledge and assessment techniquesand implementing of the check list.
source: Road safety in the scheme - inspection and audit of plans. 1996
Australia:A road safety audit is a formal examination of an existing or future road or traffic project,or any project which interacts with road users, in which an independent, qualifiedexaminer reports on the project’s accident potential and safety performance.
source: Road Safety Audit 1994
New Zealand:A formalised process to identify potential safety problems for road users and others, andto ensure that measures to eliminate or reduce the problems are considered fully. A safetyproblem is defined as a feature which has been identified from a drivers perspective whichgives a misleading or confusing message.
source: Safety audit policy and procedures 1993
StatusOnly in the UK is the RSA mandatory for all trunk roads. In New Zealand, a RSA is mandatoryon a 20% sample of new state highway projects. The status on other type of roads and in theother countries is ‘recommended’.
Mandatory RSA
United Kingdom Norway Denmark Australia New Zealand
7 7 (20% sample of state highways)
Types of roadThere are different types of road for which a RSA is issued. These types are Rural (motorways,express roads or trunk roads) and Urban roads. Another differentiating element is whether aproposed scheme of an existing road undergoing a RSA is known or unknown to the Audit
24
Team. When the scheme subject to auditing is part of a periodical (or even annual) maintenanceprogram it is more likely that the scheme is ‘known’ to the Audit Team. The Norwegian AT tobe formed will probably consist of a group of county officials auditing a scheme from one of itsmembers.
Type of Road subject to be audited
United Kingdom Denmark Norway Australia New Zealand
Motorway, Trunkroad or Urban road
Rural Urban Rural Urban MotorTrunk
Rural Urban Rural Urban
Known orUnknown
Known andUnknown
Unknown (Known) AT shouldhave ‘fresheyes’
AT should be ‘notregular users’ in thecase of a stage 5 audit
ProceduresThe initiator of a RSA project, the one who ‘starts’ the Audit, as well as the one who is finallyresponsible, i.e. decides whether or not to implement solutions to ‘the remarks’ made by theAudit Team, are mentioned below. The number in this table refers to the paragraph whichdescribes the procedure of the Audit. Also added in this table is, whether the results of a RSAare open to the general public. Norway and Denmark have not discussed the item in depth yet.In Denmark a general rule exists providing the general public access to documents produced inor at behalf of the public administration. In the UK and Australia RSAs could be used in courtcases. In New Zealand some audits become evidence in so called planning courts who decidewhether a (usually contentious) project will proceed. The last row in this table indicateswhether there is discussion between the auditors and designers regarding the RSA results.
Initiator, Person finally responsible, Organisation, Public Access, and Discussion.
United Kingdom Denmark Norway Australia New Zealand
Public Access available at Public Inquiries* Yes Pilot notaccessible
available atPublic Inquiries
some
Discussion Dep.of Transport Possible ** Possible Not necessary Yes, withDesigner orClient
Not intended
consultant sometimes
Various Counties Yes
* source: HA 42/94 Vol.5 Sec.2.2 ** source: HA 42/94 Vol.5 Sec.2.29-30.
The Audit Team, quantifications
25
Independence is important in road safety auditing. Yet there are many different ways to ensurethis independence, and even within some countries there are differences. An Audit Team couldfor example come from inside or outside the ‘designer’s organisation’. The capability and thenumber of people in an AT also differ from project to project. The AT can even containtrainees. The qualifications / experience of the audit team do not vary much between thecountries. It is has been found that a RSA could be carried out by one person, but in mostcases more personnel is recommended. The New Zealand’s audits of existing roads tend to bequite large; four members.
Qualifications of the auditorExperience in design and implementation, designing remedial measures, and knowledge ofaccident figures, are highly desirable. Furthermore, the auditor should have knowledge of roadsafety in general, design standards, road user behaviour and road safety measures in general.Experience with accident investigation techniques is a must for an auditor. It is also importantto know what is going on in the designer’s world, and stick to known practice. The police assistance is used in stage 3, thus most of the time only minor modifications aremade by them. It is preferable to ask them to join in at stage 2, but this seems to be difficult toarrange. Training courses are necessary to pick up problems, but not enough on their own; one
also needs experience. Two members of staff can also join in the team and one person alwaysvisits the site at daytime and at night .
Independence, Size, and Qualifications.
UnitedKingdom
DoT: outside the organisation, other cases: lessstrict.
Independent, but the could have a staff memberinvolved in the auditing
1 - .. Different skills and experience
Other elements with respect to the auditing team are:0 The team should include specialised safety engineers with experience in accident
investigation and analysis.0 In order to ensure that the procedure is as objective as possible, the auditors should be
independent of the design team. This is insisted on by the Department of Transport.0 Attention should be paid to all road users: pedestrians, (especially children), bus drivers
and passengers, cyclists as well as motorists, especially for urban schemes, and theirneeds should be considered. In order to achieve this, the auditors should take the role ofall users and try to predict/visualise, as precisely as possible, the way different users willperceive the scheme (‘drive, ride, walk’ concept).
26
0 Consultation with experts outside the auditing team (such as traffic signals engineers, thePolice) may be necessary.
ManualsEvery country uses manuals of their own, nevertheless all designers of manuals have looked atthe UK manual. Except for the UK Department of Transport regulations and the TNZprocedures conducted on a 20% sample on state high ways, they are guidelines and do nothave to be obeyed unquestionably.
Manuals used
UnitedKingdom
Dept.of Transport Compulsory RSAs
I.H.T. Guidelines for the Safety of Highways
Various Counties Various manuals
Denmark Road Directorate Manual of Road Safety Audit
Norway Public Roads Administration Guidelines for inspection and Quality Audit
Australia Austroads Principles and advice on good practice
New Zealand Transit New Zealand Guidelines to policy and procedures
UtilizationWhen conducting a RSA, the audit team should not try to redesign the scheme, instead theyshould pay attention to road safety for all kinds of different road users, and their suspected roaduser behaviour. The way this should not be done, is to compare the design with relevantstandards and see if it matches, but the audit team should check if the design appropriatelyinteracts with the design standards, for strictly applying standards does not always lead to asafe road.Some other findings about RSA are mentioned below. It is important that a site visit is carriedout. Both in daylight and at night. Thus the visibility for different road users can be checked inthe context of the road and its surroundings. When a RSA is carried out in an early stage of thedesign process it is less likely that ‘errors’ become embedded in the design and become harderto correct later on. A RSA should not seriously delay a design process, thus attention should bepaid to the embedding of the RSA during the planning of the design process. Attention shouldbe paid to monitoring and feed back to the audit team after opening of the road when accidentsoccur.The RSA process should be formally organised and its outcomes documented. Concerning theformalization and purity of an audit, it is to be recommended that the audit results aredocumented before there is discussion (if any) with the client, and concessions could arise.Some say that a formalised RSA leads towards a more systematical approach and enlarges thechance of a consistent outcome. The ultimate grade of formalisation is to make a RSAmandatory. Relevant plans and documents should be available to the audit team and should bementioned in the audit report. It should be clear what should be audited, which tasks there are,and who is responsible for those tasks. It can be beneficial to use the same names and numbersof stages for less misunderstandings and for a better comparison with other RSA documents.
Time spent by auditors
27
The time spent on conducting a RSA varies considerably. The responding consultants in aBritish survey needed 8 to 109.3 hours with an average of 58 hours to complete a RSA. TheRSA undertaken by the local authorities themselves took from 2 to 80 hours, with an averageof 21 hours. The overall weighted average for completing an audit was found to be 25 hours.Time could be reduced in the later stages if a feasibility stage RSA was carried out and therewere informal consultations throughout the design of the scheme.
Redesign CostsThe following table indicates the average percentage of construction cost as a result of redesigndue to the audit, compared to the original construction cost for varying sizes of schemes. Thecombined average percentage for audits carried out by both local authorities and consultants,was found to be 0.72 %. Smaller schemes tend to require a greater percentage. In some casesthe audit team recommendations even led to a cost reduction.
Proposal for the development of a frameworkA framework for the development of RSAs can be found in five points containing tasks for thedifferent bodies responsible for different aspects of road safety.0 National governments should develop RSA procedures and methods.0 National road authorities should perform pilot audits for all roads, including TERN.0 National organisations responsible for design guidelines & manuals should integrate RSA
in tools for improving road safety.0 National road research institutes should evaluate RSA.0 European Commission should initiate pilot audits on TERN roads. These pilots should
point out how the audits for TERN roads will be performed, with regard to all previouslyrecognized ‘levels’; procedures, the audit team, and responsibilities.
Introducing RSAProbably the best way of introducing RSA is ‘top-down’ (management and governmental)approval and ‘bottom-up’ (road designers) training. In this introduction stage the use ofchecklists could be useful. When introducing RSA, knowledge of accident investigationtechniques or safety engineering is necessary. Another crucial point in introducing RSA is tohow to tell when a designer is wrong. The best answer to this problem is probably an increasein accidents. On a European level, the procedure could perhaps be used in a highly aggregatelevel, using local knowledge of road safety when performing a RSA.
General conclusions
28
0 Some European countries have developed a national RSA system. Many people involvedin this development think of RSA as a promising way to improve road safety.
0 RSA Pilot projects point out that design inaccuracies can be discovered in new roaddesigns and RSA evaluations already carried out in some counties have been verypositive; RSA seems to work.
0 The introduction of a RSA system can either be done bottom-up or top-down. A bottomup approach can lead to a vast, enthusiastic participation, whereas a top-down approachcan lead to a more explicit introduction.
0 Although road safety comprising design solutions can be tracked using a RSA, theprecise effects are yet still unknown.
0 Quality is added to a national road system by using a RSA system.
1.3 The phenomenon Express road
Workpackage 3 dealt with express roads, in the Technical Annex of the SAFESTAR projectprovisionally defined as “[...] roads which are between the well known and well definedmotorways on the one hand and ordinary single carriageway rural roads on the other.” Aninventory in a number of EU countries showed that this type of ‘intermediate roads’ exist inmost countries, although generally named differently. Express-type roads are often found inmore than one category of the national road classification systems. In order to allow work toprogress, more clarity was needed on what exactly an express road is. On the functionalcharacteristics of express roads there appeared to be a high level of agreement in Europe. Based on the outcomes of an expert workshop on the issue and based on the inventory thefollowing functional definition emerged (Van Schagen & Hummel, 1998):
An express road is a high capacity road for long distance traffic with limited access andclosed for non-motorised traffic.
The last part of the definition (‘closed for non-motorised traffic’) means that express roads donot exist in the UK and Ireland, since in these countries all non-motorway roads are all-purposeroads, i.e. open to all traffic. In Sweden and Portugal a limited number of roads which wouldotherwise fall within the category of express roads, cannot be classified as such, because theyare also open to non-motorised traffic.
The design characteristics of express roads differ widely both between countries and withincountries in particular with respect to cross-sectional design (single and dual carriagewaydesigns) and intersection design (at-grade or grade-separated). Therefore, it was impossible tocome to a geometric definition.
From a detailed accident analysis on the Portuguese situation and from available data of othercountries (Cardoso & Costa, 1998), it becomes clear that express roads have a bad safetyrecord when compared to roads with a full motorway design. Compared to ordinary roads thesituation is more complex. Whereas in terms of (injury?) accident rates express roads performbetter than ordinary roads, the death rates are very similar, indicating that accidents on expressroads generally result in more severe injury. Table 1.1 provides the data for the Portugueseroad network.
29
Like on motorways, run-off-the-road accidents are the most frequent accident type on expressroads. On ordinary roads lateral accidents are the most frequent accident type. Lateralaccidents are the second most common accident type on single carriageway express roads, rear-end accidents on dual carriageway express roads and motorways. Table 1.2 presents the dataon accident types as found in the Portuguese accident study by Cardoso and Costa (1998).
FATALITIESDEATH RATES
(per 106 Vehicle km)ACCIDENTS
ACCIDENT RATES(per 106 Vehicle km)
1990 1995 1990 1995 1990 1995 1990 1995
All roads 1298 1205 0.068 0.043 14478 15940 0.761 0.569
Table 1.1 Numbers and rates of fatalities and injury accidents on different types of roads inPortugal in 1990 and 1995.
It is clear that for safety reasons it would be better not to build new express roads and toupgrade all existing ones to motorways. However, the decision on the type of road appears tobe based mainly on (expected) traffic volume and financial resources. Occasionally,environmental and land use considerations play a role as well. Safety arguments, on the otherhand, do not seem to play an important role in the decision making process. Although it seemsjustified to state that the vast majority of decision makers are well aware of the fact that from asafety point of view express roads are not a satisfactory option, other arguments seem tooutweigh the safety arguments.
COLLISION HITPEDESTRIANS
RUN-OFF--THE-ROAD
SUMFrontal Rear End Lateral Obstacle
2x2 Motorway 2 25 8 5 4 47 91
2x2 Express road 7 25 14 5 8 30 89
2x2 Ordinary road 5 15 24 14 23 16 97
2x1 Express road 16 16 25 4 8 29 88
2x1 Ordinary road 20 11 29 4 11 20 94
Table 1.2 Distribution of the (injury?) accidents in different road classes by accident type(percentages).
Even if safety arguments would get a higher priority in the decision making process, e.g byapplying strictly the one million ECU rule (saving one life justifies the investment of one million
30
ECU), there will still be situations where a motorway is not a realistic option, for example if the(expected) traffic volumes are low. If express roads are the only realistic solution, it must bemade sure that their design is as safe as possible. In the next sections the recommendations forhorizontal alignment, cross-section, safety devices (including road side safety) and intersectionsat express roads are discussed. Since experimental research on express roads is virtually non-existent, the majority of recommendations are mainly based on two-lane rural highways andinterurban roads.
31
2. ALIGNMENT
2.1 Horizontal (and vertical) alignment
2.1.1 Express roads
From a safety point of view, curves are a critical design element in the horizontal alignment:According to Zegeer et al. (1992) accident rates are 1.5 to 4 times higher in curves than ontangents.In Portugal, around 20 per cent of the accidents on dual carriageway roads (all categories) and25 per cent of the accidents on single carriageway roads (all categories) happen in curves. Thenarrower the curve, the higher the accident rate (Hughes et al., 1997). Accident rates areparticularly high in isolated curves, in the first of a series of curves or in narrow curve followinga number of relatively wide curves (OECD, in preparation). Hence, for safety, the location anddesign of curves is of importance as well as the design consistency between curves.
The fact that accident rates in curves are higher than on tangent sections does not mean thatcurves should be avoided altogether. The use of straight sections longer than 5 km. is generallydiscouraged because of the risk of drivers becoming drowsy and less alert. In general, it can bestated that the design speed should be guaranteed throughout the entire road, including atcurves. The following formula for determining safe and comfortable curve radii isrecommended (Hummel, 1998):
Rf i
Vh
z
≥ +127 100
20
( / )
Rh = curve radius in m.V0 = design speed in km/hfz = side friction coefficienti = superelevation in %
For determining the minimum radius for a given design speed, the side friction coefficient fz isgiven in the following table. The friction coefficients fz in this table are based on driving-comfort.
V0 in km/h
50 70 90 120
fz 0.18 0.16 0.13 0.1
Most studies find a positive safety effect of transition curves, although some studies (e.g.O’Cinneide, 1995) report negative effects, possibly because drivers underestimate thesubsequent curves. Hummel (1998) concludes that the use of transition curves on high speedrural roads (express roads) must be recommended for safety reasons. Superelevation also has apositive effect on safety, though it should not exceed 8%.
32
With respect to vertical alignment it can be concluded that safety is aversely affected whengradients increase: a slight increase in accident rates with gradients up to 6 or 7 per cent and alarge increase beyond those percentages (Slop et al., 1996). The accident rates for downgradesis much higher than for upgrades (Zegeer et al., 1992; Slop et al., 1996). When the horizontalalignment is relatively bendy, the negative effect of steep gradient increases.
A specific effect of vertical alignment is the blind effect caused by vertical convex curves,affecting the sight distance. This seems to be of particular importance for single carriagewayexpress roads. Slop et al. (1996) present the following formula for determining the radius of aconvex vertical curve:
RS
h hv
a
e
0 5 2
02
,
( )+
Rv = radius of vertical curve (convex)Sa = actual sight distancehe = eye heightho = object height
The value to be substituted for Sa depends on what the designer wishes to offer to the roaduser. The minimum requirement is the stopping sight distance.
Regarding concave (sag) curves a minimum radius of 3,000 m. is recommended in order toavoid insufficient sight distance when using dipped headlights at night.
2.1.2 Rural roads
This report uses design consistency or relation design as the leading design principle for thehorizontal alignment. Successive straight and curved road sections should be related to eachother in such a way that road users will not be subjected to unexpected and uncontrollablechanges in the horizontal alignement.The procedure for attaining a consistent design is based on both qualitative models which relatedesign elements to the operating speed (V85), and a comparison between the calculatedoperating speed and the design speed.
Tangents and curvesBoth design consistency and relation design are an important part of the design philosophy inthis report. However, it is a fairly theoretical principle from the (accident) researchers’ point ofview. The relationship between consistency characteristics and road safety indicators is still thesubject of research.One of the assumptions in this respect is the influence of the length of a tangent on the speed inthe curve. Lamm et al. (1988) considered this as an essential element in their methodology forrelation design. But Fink & Krammes (1995) showed that this influence is rather small or non-existent regarding the relationship between accident rate, degree of curve and length of tangent.The influence of the preceding tangent is mainly concentrated in the speed at the end of thetangent and the speed reduction when entering the curve. Cardoso (1996) related these twotypes of speed to the accident rate in both curve and tangent (lane width and AADT are alsopart of this relationship).
33
V85 34.7001.005DC�2.081LW�0.174SW�0.0004AADT (1)
Instead of using speed indicators, one can use a less direct indicator for the safety of the road.This indicator is the work load of a (car) driver. It is supposed to be related to the complexityof the road and traffic environment. Messer et al. (1981) have developed a methodology fordetermining the work load of a driver on a given road. The work load is determined by scoringthe level of complexity of different road and traffic elements of a given road section, of thepreceding, and of the following road sections.However, this methodology proved to be very dependent on expert opinions.
Krammes et al. (1995) have introduced an alternative method: the vision occlusion method. Inthis method drivers are asked to drive with their eyes closed until they think it is necessary toopen them again. The time period they drive with their eyes open is considered to be anindicator for the work load. The work load is:0 WL = 0.193 + 0.016*D for curves, D is ‘degree of curve’ and R² = 0.90;0 WL = 0.176 for tangents.The value for tangents implies that drivers drive with their eyes open in only 17,6% of the timethey are driving on tangents.
Operating speedIn general operating speed and road class are highly correlated, e.g. main roads have beendesigned in such a way that high speeds are possible and acceptable. However, many localcircumstances will temporarily or permanently influence the operating speeds, like bendiness,density of traffic, crossing traffic, atmospheric conditions, road width etc. Road design dealsmainly with conditions which are permanently present and which are more or less ideal:vehicles are in a free flow, only influenced by infrastructural elements. These type of conditionshave been related to operating speed by means of statistical relationships (models): Theoperating speed on a cross-section (spot) is related to the characteristics of that cross-sectionand of the road section. This spot speed approach (one cross-section) seems, besides theoriginal question, to find a relationship between horizontal alignment and the speed along theroad section (speed profile). Spot speed is not quite the same as the speed profile (horizontallyvarying) of a road section. But a speed profile can be ‘constructed’ by choosing the spots atimportant road features which make the speed change substantially. The most dominatingfeature in this respect is a curve. Many relationships between operating speed and curvecharacteristics have been developed; Cardoso et al. (1997) have given an overview. Some ofthese relationships are given below in order to show the type of relationships:
Lamm & Choueiri (1987) derive V85 from the relationship:
R²=0.842
V85: 85-percentile speed in mph
1) The relationship between the radius of a curve (R) and the degree of curve (DC) is: R=5730/DC(R in feet) or R=1746.4/DC (R in m).
34
V8558.6561.135DC (2)
V85 102.4 1.57DC � 0.012L 0.10� (3)
DC: ‘degree of curve’ (0o tot 27o)1
LW: lane width in ft.SW: shoulder width in ft.AADT: annual average daily traffic (400 to 5000 motorvehicles per day)
A more simple relationship is given by the same authors:
R²=0.787
This more simple model (2) is also available for lane widths of 10, 11 and 12 ft.
Collins & Krammes (1996) have found the following relationship:
R²=0.82
V85: 85-percentile speed in km/hDC: ‘degree of curve’L: curve lengte �: ‘deflection angle’ (=L*D/100
Results from SAFESTAR The relationship between operating speed V85 and road characteristics was modelled for bothcurves and tangents. The following characteristics of the curves were available:� curve radius (in m)� curve length (in m)� lane width (in m)� shoulder width (in m)� longitudinal gradient (in m per 100m)
Speed models for curves were fitted using data from Greece, Finland, France, and Portugal.The number of available curves was 5 from Finland, 30 from France, 9 from Greece, and 36from Portugal. Speed measurements added up to 10,000 vehicle speeds in each country.
Separate models were fitted for each country. A general model, using the data of the fourcountries, was also fitted. However this general model still needs a (constant) factor for explaining a part of the ‘national’ statistical variance. Worse is that this general model is notstatistically significant regarding this constant factor. A general model which was independentfrom the national differences could not be fitted. So only the four separate models remain to beused. These for national models for curves are:
35
Finland: V85 � 51.756 � 337.780 / R � 0.6049 � V85, AT (4)
France: V85 � 49.220 � 292.736 / R 2� 0.454 � V85, AT (5)
Greece: V85 � 41.363 � 294.000 / R � 0.699 � V85, AT (6)
Portugal: V85 � 25.010 � 271.500 / R � 0.877 � V85, AT (7)
R² = 0.707
R² = 0.801
R² = 0.916
R² = 0.896
V85: 85 percentile speed (in km/h)R: curve radius (in m)V85,AT: 85 percentile of the unimpeded speed on the tangent preceding the curve (in km/h)
Figure 2.1 shows these relationship graphically for an approach speed of 90 km/h. Thedifferences between the four countries are rather small, except for the curve speeds at smallradii in France.
TangentsSpeed models for tangents were also fitted separately for each country, because a generalmodel could not be fitted. Many different characteristics of the tangents were used in fitting theequations:� average bendiness (total deflection angle along the tangent; in degrees per m)� average lane width (in m)� average shoulder width (in m)� total percentage upgrade� total percentage downgrade� average gradient (in m/km)� total hilliness (sum of elevation change, both upgrade and downgrade; in m/km)The number of available tangents was equal to the number of curves: 5 from Finland, 30 fromFrance, 9 from Greece, and 36 from Portugal.
The selected characteristics are the outcome of a statistical optimization process. The literaturereview did not reveal theroretical models which give a sound basis for a ‘true’ relationshipbetween tangent speed and road characteristics. National differences concerning topography,land use, road design, and road user behaviour appear to be an important factor for the
resulting set of road characteristics. So, for each country, a different set of road characteristicswas found to be related to tangent speed.
Figure 2.1 Unimpeded speed at a curve related to curve radius, using data from four differentcountries
R² = 0.768
R² = 0.653
R² = 0.918
R² = 0.821
V85: 85 percentile of the unimpeded speed (in km/h)L: tangent length (in m)Bend: bendiness (in degree/km)LW: lane width (in m)Hill: hilliness (in %)Prad: curve radius of the curve preceding the tangent section (in m)
37
France: AR �
e 21.98� AADT 0.5745
� MaxRedV 0.351885
MaxV 0.5.01785
(12)
France: AR �
AADT 0.2536� WL 5.536
� MaxAppV 3.03185
e 4.708� LW 2.355
(13)
Portugal: AR �
AADT 0.4174� MaxRedV 0.008172
85
e 0.4284(14)
Portugal: AR �
AADT 0.4065� WL 0.6420
� MaxAppV 0.637085
e 2.582(15)
Results from SAFESTARAccident rate models were fitted for both types of input: speed indicators and work load. Theroad characteristics which were used for these models were:� average daily traffic (all motorized vehicles in both directions)� maximum estimated spot speed on each road element (both directions)� maximum estimated approach speed on each curve element (both directions)� maximum estimated speed reduction from preceding element (both directions)� sum of estimated speed reductions form preceding element (both directions)� curve radius� length� length of preceding tangent (in case of a curve)� shoulder width� grade (longitudinal gradient) of a curve
The database consisted of 1,000 road elements, with a total length of 611.7 km, in threecountries, Finland, France, and Portugal. However, the number of accidents on the Finnishroads appeared to be too low for fitting equations to data (many elements with no accidents atall). So it was decided to fit the models with the data from France and Portugal (269.5 km roadlength).The curve models have been fitted for France and Portugal separately:
R² = 0.58
R² = 0.54
R² = 0.20
R² = 0.24
38
AR �
AADT 0.4057� MaxRedV 0.09947
85 � MaxAppV 0.655685
e 1.839� LW 1.14
(16)
AR �
AADT 0.316� WL 1.593
� MaxAppV 2.43385
e 6.957� LW 2.034
(17)
France: AR �e 13.72
� AADT 0.1638
MaxV 2.88185 � LW 0.2528 (18)
Portugal: AR �e 0.9453
� AADT 0.5246
MaxV 0.0628185 � LW 0.7000 (19)
AR: accident rate (number of accidents per million vehicle miles)AADT: (avarage daily traffic)MaxRedV85: maximum estimated speed reduction from preceding element (both directions)MaxAppV85: maximum estimated approach speed on each curve element (both directions)MaxV85: maximum estimated spot speed on each road element (both directions)WL: work load (Krammes et al., 1995)
Obviously, the French models fit better than the Portugese models. The databases of both countries were merged in order to fit more general curve models. Thisresulted in models with rather low R² values:
R² = 0.13
R² = 0.18
The accident rate on curves according to equation (16) has been put into a graphicalrepresentation (Figure 2.2). The accident rate in Figure 2.2 was calculated with an approachspeed of 100 km/h and an AADT of 7,000 motor vehicles per day.
The modelling for tangent elements resulted in two national models:
R² = 0.09
R² = 0.33
AR: accident rate (number of accidents per million vehicle miles)AADT: (avarage daily traffic)MaxV85: maximum estimated spot speed on each road element (both directions)LW: lane width (in m)
39
Figure 2.2 The relationship between accident rate (AR) and speed reduction(MaxRedV85) for different lane widths (LW)
2.2 Marking and Signing in Curves
The effects of different signing and marking principles have been studied by the Danish RoadDirectorate, in cooperation with SETRA-CSTR and CETE Normandie Centre in France. TheDutch institute TNO has also contributed with a simulation test.In this workpackage, a framework is developed to classify substandard horizontal curves onrural roads into a number of danger categories. A basic signing and marking concept, whichshould be the minimum applied to curves belonging to the relevant category, is associated witheach of the danger categories.By the use of both simulation and full scale tests, the effectiveness of the proposed signing andmarking strategies is tested.
2.2.1 Current practice
Since the early days of motorised traffic, signs have been used to warn drivers that they areapproaching a substandard curve. To establish a safe and comfortable journey throughsubstandard curves, advisory speed signs, curve warning signs, background markings, andmarkings on the road give an indication of the desirable speed.Current practice for the use of signing and marking of substandard curves differs widely. Evenwithin individual EU-member states, there proved to be considerable variations in actualsigning and marking of identical substandard curves. There seem to be no clear relationbetween the actual ‘danger category’ of a substandard curve and the applied signing andmarking of that curve. Because road-users do not receive consistent information on theapproach of a substandard curve, the choice of correct speed and driving behaviour is oftendifficult.
40
Safe and efficient traffic behaviour is greatly influenced by the geometrical features of the road.A review of accident-spot maps show that accidents in rural areas tend to cluster on curves,particularly on very sharp curves. In Denmark, 20% of all personal injury accidents occur onrural road curves, and 13% of all fatalities in traffic accidents occur at horizontal curves in ruralareas. The situation in France is even worse. Here 21% of all fatalities occur in curves on ruralroads. A similar situation prevails in the rest of Europe.Curve accidents are mainly caused by improper (too high) speeds. Many of those speed errorscan be related to inconsistencies in the horizontal alignment which surprises drivers by suddenchanges in road characteristics. As a result they exceed the critical speed of a curve, leading toloss of control over their vehicle.It can be very difficult for road users to perceive horizontal curves, and it is almost impossibleto estimate the design speed before entering the curve. Serious safety problems occur insituations where large differences exist between the operating speed on the upstream horizontalalignment (the approach speed) and the design speed of the substandard curve. Two mainapproaches can be distinguished for solving this problem:- make the design speed of the curve similar to the approach speed;- use marking and signing of curves to provide the driver with the information needed for
correct and timely estimation of the design speed.
For financial reasons the former approach is often not possible. In most cases signing andmarking is therefore the only remaining approach for solving safety problems on curves.
2.2.2 Conceptual approach
In order to establish a more uniform use of signing and marking and thereby increasing roadsafety, a framework for signing and marking of substandard curves is developed. The purposeof the framework is to create a tool for road authorities to classify curves on rural roads into anumber of danger categories in a structural and uniform way. When a unique and uniformsigning and marking strategy is used for each danger category, the choice of correct speed anddriving behaviour is simplified, resulting in an increase in safety on curves.The input parameters of this approach are the approach speed on the horizontal alignment andthe design speed of the curve. When the input values are known, the danger category of theroad curve is determined by a model. The determined danger category of the curveautomatically results in a recommended signing and marking strategy for that curve.
Approach speed:The final signing and marking plan on each curve should be based on real speed measurements.For any overall classification of a large number of curves, the following formula for estimatingthe approach speed on two-lane rural roads has been developed:
V a = 0 .07716 * V + 2 * 0 .8 * (L - 100 )22 n - 1
Va: approach speed (km/h.) Vn-1: 85th percentile speed of the previous curve (m/s)L: distance between the present curve and the previous curve (m)
41
Depending on the type of road Vn-1 is calculated with one of the following formulas:2-lane and 3-lane roads with minimum width of 6 m.: Vn-1 = 102/(1 + 346/R1.5)2-lane roads with a width less than 6 m.: Vn-1 = 92/(1 + 346/R1.5)
Design speed of curveThe design speed of a curve is the speed at which a passenger car can drive, in a safe,controlled, and reasonably comfortable manner, through a curve under normal weather androad conditions when the road surface is wet.Estimation of the design speed is done with the aid of models or formulas. A simple butinternationally used and accepted formula for estimating curve design speed is the fundamentalcurve-design equation:
V d = R * g * (e + F )
Vd: curve design speed (m/s)R: radius of curve (m)g: acceleration due to gravity (9.81 m/s2)f: side friction factore: superelevation
The side friction factor depends on the condition of the road surface, the tyres of the vehicleand the vehicle speed. Most national guidelines for road construction table include averagevalues for the side friction factor. The following table shows average values for the side frictionfactor as used in Denmark.
Classification of substandard horizontal curvesWhen the approach speed on the horizontal alignment and the design speed of the curve areknown, the developed model can be used to determine whether a curve is to be classified as‘substandard’. In the case a curve is substandard, the danger category in which the curve can beclassified, can be determined with the same model.The model is calibrated for the Danish road network. The authors stress that the model canonly be used outside Denmark after calibration for the prevailing national conditions.
Curves with a design speed higher than, or equal to the approach speed are not regarded assubstandard curves. Signing and marking is therefore not required on these curves.
42
Figure 2.3 Model for classifying horizontal substandard curves into danger categories
Basic signing and markingBased on existing literature, several national signing and marking guidelines in Europe, expert-meetings, a simulator test and full-scale tests in Denmark and France, the following conceptsfor singing and marking of curves have been developed.
Each basic signing and marking concept consists of one or more of the five elements:- delineators- centre line and edge lines- advance warning- advisory speed signs- chevron signs.
43
Figure 2.4 Basic signing and marking for each danger category
2.2.3 Evaluation of new types of marking
Centre and edge lines should be used at all substandard curves. On curves in the dangercategories A, B, and C, the lines can either be ordinary painted/thermoplastic lines or profiledlines. Based on research in Denmark and other countries, profiled edge and centre lines arerecommended for curves belonging to danger category D or E.In this study, the effects of profiled road markings were studied on four test curves inDenmark. On the test curves ‘vibracomb’ lines were used as edge and centre markings. Thesevibracomb lines consist of an ordinary line to which a profiled line is attached, like the teeth of acomb. The teeth of the comb are pointing towards the vehicle lanes.Visibility of these profiled markings proved good, because of their height. Moreover, an audiblerumbling effect occurs when a rotating tyre touches them, thus giving the driver an audiblerumbling warning.
44
Figure 2.5 Vibracomb line
Although tendencies towards speed reduction were observed on the curves with vibracomblines, the observed changes were not significant.The number of vehicles infringing the centre line proved to decrease on the curves withvibracomb centre lines.These positive effects are supported by other research studies on profiled markings in Europeand the USA. Here profiled lines also improved driving behaviour and reduced infringements ofboth centre lines and edge lines. Accident studies showed a reduction of the number of run-offaccidents by 60% - 70% on curves with profiled edge lines.
Tests with road studs, as conducted in this study, showed less positive results. The test showedthat the costs of applying the road studs were high, and that their retro-reflective capacity wasreduced to almost zero within a few days, because the self-cleaning function did not operate asexpected. Furthermore, most of the road studs were destroyed after only one or two months,due to winter maintenance.
2.2.4 Recommendations for road design
Simulator tests and full scale tests proved the positive effects of the developed signing andmarking strategies for the different danger categories. Additional positive effects due to themore uniform signing and marking strategies and the improvement on the recognition ofdangerous curves, add to the value of the developed signing and marking strategy.The use of this structural approach should therefore be recommended.Detailed recommendations on when and how to apply signs and markings are given by Nielsenet al. (1998).
The study on signing and marking on curves in rural roads also gives a number of measures thatcan be added to the basic signing and marking of curves where special road safety problemsoccur.
45
Poor visibility of the curveVegetation: Vegetation on the inner side of a curve makes the curve less visible and
should be reduced. Vegetation, such as shrubs and small trees on theouter side of the curve, will on the other hand improve the visibility ofthe curve and also give good guidance through the curve. This isespecially valid during daylight.
Pavement marking: In some cases painted signs on the road surface can improve thealertness of drivers.
Poor readability of curveVegetation: Vegetation on the inner side of a curve makes the curve less visible and
should be reduced. Vegetation, such as shrubs and small trees on theouter side of the curve, improves perception of the direction and thesharpness of the curve in daylight.
High speedsPerceptual illusions: Special road markings may help the driver to choose the most adequate
speed at places where the accident risk is often underestimated.Electronic speed sign: Electronic advisory speed signs which show the curve design speed,
and flash when a vehicle exceeds this speed, can help estimating thecorrect speed.
Improved road surface: A rough road surface, which increases the noise level in the vehicle, ora surface in a different colour can also cause drivers to choose a lowerspeed.
Pavement marking: In some cases painted signs on the road surface can improve attention.Humps: The use of speed humps in the approach can effectively reduce speeds
in the curve.Carriageway width: Reduced carriageway width on the approach and on the curve itself can
reduce speed.
High frequency of run-off-the-road accidentsHard shoulder: Improvement of the verge e.g. by making a paved hard shoulder, can
reduce run-off-the-road accidents.Safety barriers: Safety barriers do not reduce the number of run-off-the-road accidents,
but can reduce the severity of these accidents.Road surface: Improvement of the friction of the road surface can reduce the number
of run-off-the-road accidents
High frequency of head-on collisionsGhost island: A central hatched island increases the distance between opposing
vehicle and thus reduces the risk of head-on collisions. Doublecontinuous lines can give comparable results.
Hard shoulder: A possible cause for head-on collisions is the overcompensation afterrunning off the road. These accident types can be reduced by thepresence of a hard shoulder.
46
3 CROSS-SECTION
3.1 Motorways: specific safety measures for emergency lanes and shoulders
Accident statistics of several European countries indicate that a sizeable proportion of accidentson motorways is related to emergency lanes. The cause of these accidents seems to beinappropriate use of the emergency lane and the nearside lane, for instance: vehicles avoidingruts in the road surface by partially driving on the emergency lane. This study is aimed to provide an analysis of accidents that are related to the use of emergencylanes in different European countries and subsequently produce an accident typology. Thistypology will then be used to derive possible countermeasures to prevent these accidents. Thenature of this task is an exploratory one. The results could be used as a starting point fordiscussion with road authorities in the TERN-framework.The work has consisted of five steps:1. A general literature review to compile existing standards and policies.2. An analysis of relevant accidents, using as example the European databases3. A behavioural study of road users, specifically with respect to the inappropriate use of
emergency lanes. This study will primarily consist of a literature review, supplemented byfield observations.
4. Summarizing and interpreting the results and producing recommendations for practicalcountermeasures. Attention will also be paid to the different conditions on bridges and intunnels.
The target groups are supranational bodies and national authorities in the EU countriesresponsible for the safety of road infrastructures.
3.1.1 Literature review
At first a study of the literature was carried out. Additional standards and practices aboutemergency lanes were collected by means of sending questionnaires to all of the Europeancountries, and correspondence and interviews with colleagues of European research institutes. By means of the questionnaires, accident data of the different countries was also requested.Owing to the poor response to this item, the main national road safety research centres wereasked to deliver the data on multiple and single vehicle accidents on emergency lanes ofmotorways.
Survey of international standards and policiesThe development of motorway standards is coordinated by the Motorway Working Group(MWG) of the Directorate General DGVII Transport. EFTA countries were invited to join theMWG, and further contacts with Central and Eastern European Countries (CEES) have beendeveloped for making agreements.The agreements about emergency lanes contain a few recommendations concerning theemergency lanes of motorways and a few operational regulations. These recommendations donot have a mandatory status.
47
In order to harmonize the National standards, the START-report ‘Road Typology in theTERN’ (1994) proposed a minimal number of conditions for emergency lanes of motorways.Summarized, the following international recommendations on motorways emergency lanes andhard shoulders have already been made:- a minimal width of traffic lanes on straight alignment is recommended (3.5 m);- a minimal width of hard shoulder (paved or stabilized) is recommended (3.75 m);- the shoulders should normally include a continuous emergency stopping strip (of at least
3.00m);
In order to prevent improper use of emergency lanes and to reduce the number of stopped cars,the typical facilities spacing is recommended by Motorway Working Group (MWG) of theDirectorate General, DGVII: rest areas with parking and toilets (every 20 km), service areas(every 50 to 100 km), and service and accommodation areas (every 200 km).Also the presence of emergency calling posts are recommended:- they are to be placed every two kilometres in each direction and opposite each other (in
order to avoid the perceived possible need to cross the road);- notices explaining their functions, fixed on emergency telephone boxes; - make an European leaflet on motorway use to indicate the circumstances in which hard
shoulders should, and should not, be used. Include instructions concerning emergencytelephone use and explaining the functions of the telephones.
The relevant available national standards and practices in EU countries were collected andreported by Braimaister (1998). It deals with:- Standards, guidelines concerning motorways design- Main road design characteristics - Traffic regulations on use of emergency lanes of motorways.
Design characteristics and regulationsUsing the results of the questionnaires and the additional information by E-mail, the followingdesign characteristics of emergency lanes and traffic regulations on use of emergency lanes ofmotorways were collected from the different European countries.
Main design characteristicsThe width of emergency lanes is considered as most important design characteristic.Accordingly to the recent TERN-typology (START), the width of emergency lanes on TERN-motorways should be at least 3.00 m. The present norms in most of the EU countries do notmeet this requirement. Only France (partly), the Netherlands, Portugal, and the UK havealready standards which meet this requirement; or they have greater widths. In other countries,the width of emergency lanes is mostly 2.5 m.
Emergency phonesThe presence of emergency phones on motorways is better harmonised . Almost all of theEuropean countries have such phones every 2 kilometres on motorways (in UK even every 1.5km). Finland and Sweden do not have emergency phones at all.
48
Traffic regulationsIn all the EU countries, the traffic regulations commonly prohibit use of emergency lanes forregular traffic. The purpose of emergency lanes is to give space for emergency stops of vehiclesand the use by special vehicles which belong to police, ambulance, or fire brigade.There are discrepancies in the lists of reasons, which are to be accepted as emergency cases indifferent EU countries. All the countries respect the following reasons:- police, fire brigade, or ambulance in action- road maintenance in operation- a breakdown of a vehicle- driver suddenly becoming unwell.
All the countries, with exception of Germany and Greece, respect the reason for stopping onemergency lanes in case of ‘actual shortage of fuel’. Such a ‘reason’ is to be considered as aviolation of traffic rules.‘Assistance in case of an accident’ is not allowed as reason to stop the vehicle on an emergencylane in UK and Sweden. Such a reason as ‘being at police disposal as a witness’ is legal inBelgium, Denmark, Finland, Luxembourg, Netherlands, and UK. But France, Germany,Greece, and Portugal do not accept such a reason. All the countries, with the exception ofFrance, accept ‘extremely bad weather conditions’ as sufficient reason to stop the car.
Exceptional practicesThere are also some less frequent exceptional practices in some countries: - driving escorted by police;- driving public bus on the specially marked hard shoulder during a traffic jam;- as additional traffic lane when special traffic sign is open, due to high traffic volume.
RecommendationsSince the recent TERN-typology does not have such a list of reasons, a recommendationshould be made to harmonise these lists in order to avoid dangerous situations caused bydiscrepancies in traffic regulations and usual behaviour in different countries.
3.1.2 Databases on road accidents and exposure
The international databases have been studied with the intension to estimate the accident riskon motorways and emergency lanes of motorways. For the risk calculation the following datashould be available:1. total numbers of accidents and casualties on motorways;2. types of accidents, with at least one of the involved vehicles entering or leaving the
emergency lane (hard shoulder) of motorways;3. accident data for a number of years (at least for 3-4 years).
At least the following exposure data is necessary to be able to estimate properly the relevantaccident risk:4. total length of motorways;5. percentage of motorway length with and without emergency lanes (hard shoulders);6. AADT on motorways.
49
The main available databases were consulted in order to find available data for fifteen membercountries of the EU. But not in any European database was there data available of multipleaccidents on emergency lanes. One could expect in the near future that the relevant accidentdata will be available in the European database CARE.To get the data sooner, the only possible way is by means of performing extra and expensiveresearch. Within the limited budget of SAFESTAR it was not possible to realize this research inall countries. Owing to the missing data, only an estimation could be made of the volume andrisk of multiple accidents on emergency lanes in European countries.
Only in two countries was data found from multiple injury accidents on emergency lanes andaccident rates: United Kingdom (1979-1980) and the Netherlands (1979-1982; 1992-1995). Tothis data, exposure data from IRTAD database was added.In order to estimate the risk of multiple accidents on emergency lanes on European countries,the available total accidents on motorway was collected for the year 1995 from IRTAD data.Using Dutch ratio’s an indicative estimation of these accidents and deaths in EU countries peryear was calculated.
Dutch dataIn the Netherlands multiple accidents on emergency lanes had a share of 1.5% injury accidentson Dutch motorways and 8% fatalities (1992-1995). The older Dutch and UK figures werehigher. For the injury accidents in both countries, 2.8% and for the fatalaties resp. 9.5 and10.7% was found. For the estimation of the proportion of such accidents in EU countries, ischosen for the recent Dutch figures.Using these figures, we can conclude that at least 1000 (rounded up from 967) of such injuryaccidents take place each year on all motorways in EU-countries and about 300 (rounded upfrom 280) persons are killed. If we take into consideration that the Dutch accident rates arebetter than average in Europe, we can conclude that these estimation are minimal.
Data from the United KingdomIn 1980 - 1982 a working group on accidents on hard shoulders in the United Kingdom carriedout a study to the frequency, causation, and possible means of reduction of accidents involvingvehicles using motorway hard shoulders. It was shown that the severity of these accidents wasthree times higher than the severity of other accidents on motorways.
This research resulted in three types of approaches to accident reduction on the hard shoulders:engineering, legislation, and behaviour. The group concluded the following:
As engineering measures mentioned are (advanced) road marking, surface maintenance andcable detection as a warning system for approaching drivers and the police. Owing to financiallimitations, a implementation for these measures could be based on a required minimal level oftraffic flow. In terms of cost-effectiveness there is no engineering counter-measure suitable forapplication over the whole network of roads.
Legislative measures and enforcement are considered and discussed as possible amendments tothe Motorway regulations. Only a few amendments were concluded as positive: e.g. the use offour-way hazard flashers by vehicles halted on hard shoulders. Other amendments were judgedas negative, for example: rendering obligatory immediate removal of stationary vehicles from
50
hard shoulders (suggested is that such obligatory actions could result in more delay in trafficthan stationary vehicle themselves), defining ‘emergency’ in the Motorway Regulations andeliminating the regulation which legitimizes amateur assistance. Also the compulsory use of a reflective red warning triangle is discussed; in some Europeancountries the presence and use is compulsory. The working group was against such a regulationin the United Kingdom. Placement of the triangle in accordance with the Highway Coderecommendation approx. 50-150 yards behind the vehicle, entails extra risk for the driver inwalking back down the hard shoulder.
Influencing the driver behaviour gave some positive remarks: a) fixing notices on emergencytelephone boxes and explaining their functions and b) revising the leaflet on motorway use toindicate the circumstances in which hard shoulders should, and should not, be used.The NetherlandsAccident studyIn the Netherlands, research was carried out in 1987 inspired bythe death of some break-down service officers. The research of Mathijssen (1987) consisted ofan accident and behaviour study. Included was also a literature study, but no studies werefound about the estimation of the road accidents risk caused by vehicles situated on emergencylanes (and hard shoulders) of the motorways. In relation with the accident study, at first the multiple accidents on emergency lanes weredefined as follows:- at least two (or more) road users involved in an injury accident on motorway emergency
lanes;- at least one of the involved vehicles (road users) was on, leaving, or entering the
emergency lane (hard shoulder).The number of deaths per 100 injury accidents for the multiple accidents on emergency laneswere 21.5 in contrast to the 6.3 death per 100 injury accidents of all the accidents onmotorways.
The distribution of some characteristics of these multiple accidents were:Darkness, no road lighting 25%Road works zones 6%Secondary accident (at location of primary accident) 6%Straight sections of the road 95%
Percentage of all of the multiple accidents on emergency lanes (N=171)
The distribution of some behavioural ‘causes’ of multiple accidents were:Drive too much on the right side of the carriageway 38%Accidents with pedestrians 17%Skidding of the vehicle 14%Out of control 7%Wrong way joining the traffic from emergency lane 6%
Percentage of all of the multiple accidents on emergency lanes (N=171)
Crossing the marking stripAlso in the Netherlands, observations are carried out of the frequency of crossing the markingstrip between the emergency lane and carriageway (Oldenburg, 1985). Distinction is made inthe vehicle category, carriageway lane width, and dry / wet surface.
51
The following results were found with limited observations. Vans and trucks have two timesmore cross-overs if the lane width is 3.25 m instead of 3.5 m. Also the influence of a wet or drysurface is considerable; on a wet surface there are considerably more cross-overs, both forpassenger cars as well as for vans and trucks, in comparison with a dry surface.Crossing the marking strip can be dangerous. In the study of Mathijssen (1987) was found thatabout 20% of the stopped vehicles on the emergency lanes, were less than 1 m away from themarking strip.
3.1.3 Investigations regarding vehicles stopping on emergency lanes
Vehicles stopped on the emergency lanes are potential dangerous spots. A factor to calculatethe risk for a collision with these vehicles standstill is the number of stopped vehicles onemergency lanes per 100 km motorway. So far as we know, only in the Netherlands has thisbeen investigated. The first study is already 12 years old and carried out by SWOV (Mathijssen,1987). The values, given with a contribution to the reason for the use of the emergency lanesare (both directions):
Broken-down cars + service vehicles 1.5 vehicles per 100 kmRoad work zones related 1.6 per 100 kmOthers 1.0 per 100 kmTotal 4.1 per 100 km
A second study has been carried out for SAFESTAR. The observations were performed in1997 by SWOV.A random sample of observations has shown that one had to drive an average of about 18 kmto meet a broken-down car or other vehicle/obstacle on the emergency lane at one side of themotorway. This means 10.9 vehicles/obstacles per 100 km as average at both sides. Within thisvalue, the influence of vehicles/obstacles at road work zones is large. These aspects will beconsidered in greater detail by the European ARROWS project. In comparison with the value found in 1987, the number of vehicles/obstacles per 100 kmfound in 1997 had more than doubled. The influence of the daily traffic intensity is likely to beone of the reasons.
Average waiting time in the case of a broken-down carBased on data from the road service of the Royal Dutch Tourist Association ANWB,calculations were made for the average waiting time along the motorway. The average waitingtime found for technical help was 32 minutes; the average repair time was 23 minutes. Takinginto account that the time to call technical service by using the alarm phone was about 5minutes, we get as a very rough estimation a time of 60 minutes of average stay of a broken-down car on the motorway (plus 23 minutes accompanied by the service car). These are theaverage times. Of course the amount of time depends on hour of the day, day of the week, andmonth of the year.
Warning triangle.In 1987 Mathijssen found that for only 3% of the vehicles stopped on emergency lanes, themandatory warning triangle was placed. The following reasons for this very low percentage aregiven: 1) insufficient knowledge of traffic rules, 2) absence of emergency triangle in the car, 3)trouble to get out the car and to place the triangle, 4) fear of being run over while placing the
52
triangle, 5) doubt about efficiency of the triangle as a warning device, and 6) doubt about thenecessity to warn other road users.
Work zonesThe 4th Framework Programme of DGVII started in 1997 an Advanced Research on RoadWork zone Safety Standards in Europe (ARROWS). ARROWS concerns the whole range ofmeasures of applicable road work zone safety measures, including the use of emergency lanesof motorways.
Emergency lane as an additional lane during rush-hoursIn order to obtain a better usage of the existing infrastructure, the Dutch Ministry of Transportis testing the possibilities to use the emergency lane as an additional lane during rush-hours.TNO has advised the Ministry(Theeuwes et al, 1995) concerning the necessary signalling of theexperimental sections in use.
Bus on emergency lanesAs experiment, the use of emergency lanes by buses as additional lanes during the rush-hours,was investigated in The Netherlands in early 1990. An evaluation of these experiments showthe following results:- use of short sections of emergency lanes for buses does not increase the road accident
risk when properly designed, prepared and organized;- work out a proper signalling of such location using electronic warning boards.
The preliminary recommendations to be learnt from these experiments are:- speed of buses should be reduced;- additional reserve breakdown parking places should be made available along sections of
emergency lanes;- there is no data to estimate the impact on road safety of these experiments, measured in
terms of road accidents;- according to subjective estimations of bus drivers, passengers, and other involved drivers,
the safety declined on the experimental locations.For final conclusions more tests were recommended.
Recommendations Despite the present international agreements on motorways the national standards and practicein European countries are different. The recent typology of TERN-motorways demand aharmonisation of these standards on significant parts of European motorways includedin TERN.A regular periodic check and monitoring of achievements in this harmonisation will berecommended. A randomized inventory of journey observations of a couple of hundredkilometres per country is proposed in order to produce a periodic report to the coordinatinggroup Motorway Working Group of the Directorate General, DGVII Transport.Also recommended is to gather information from the European countries about the experiencewith special use of emergency lanes, like:- during the rush-hours;- separated lane for buses;
53
- using a lane or two when the lanes in the opposite direction of the road are beingreconstructed.
This data is helpful for preparing European standards.
Only little is known about the extent and the risk of multiple accidents on emergency lanes inEuropean countries. One figure necessary to calculate the risk is the average number of stoppedvehicles and obstacles per 100 km. The value found in the Netherlands (average 11 vehicles/-obstacles per 100 km on both sides) can be helpful as reference for other European countries.
At hazardous locations with a higher risk, some additional measures can be taken:- rumble strips for marking the border between a carriageway and the emergency lane;- widening of emergency lanes;- information campaigns for road users about typical hazardous locations;- application of lighting on motorways, especially on sections where emergency lanes orcarriage lanes are narrow.
3.2. Express roads
SAFESTAR aimed at producing safety standards for express roads, with the intention ofimproving the unfavourable safety records of this road type. An extensive accident analysis,combined with an evaluation of decision making processes, expert interviews, and a literaturereview, form the basis for design recommendations.Because of the limited amount of information and knowledge of safety effects of designvariations on this road type, further research is recommended.Research for this task has been carried out by SWOV and LNEC.
3.2.1 Express roads in TERN
Motorways and single carriageway ordinary roads are two main types of roads which exist inall EU Member States and which are known by all motorists. Most countries also have somesort of roads, which do not fulfil the design criteria of a motorway, but which are of a higherorder than the ordinary single carriageway roads. These intermediate types of roads aresometimes classified under the name of ‘express roads’, but in the majority of EU MemberStates these roads do not occur as a separate category in the national road categorisation. In general, the safety records of this intermediate type of roads are bad, not only in comparisonwith motorways, but also in comparison with ordinary roads (Cardoso and Costa, 1998).
The development of TERN, the Trans-European Road Network, as set out in the 1993 Treatyof the European Union, aims to provide a road network for main international road travel,connecting all parts of the European Union. The largest part of the network consists ofexisting motorways. A substantial part, however, consists of roads which are non-motorways.Non-motorway links consist of ordinary roads and express roads. Ordinary roads generallyhave a traffic flow capacity of less than 5,000 vehicles per day. It is suggested that for trafficflow between 5,000 and 10,000/15,000 vehicles per day an express road may be the bestsolution, with the optimal traffic flow capacity depending on the exact characteristics.
54
The general characteristics of express roads are described by the Motorway Working GroupAction Start in ‘Standardisation of Typology on the Trans-European Road Network’ (1994).The Working Group recommends that express roads should:- have no urban sections;- have no private access;- do not permit parking and stopping on the carriageway;- do not permit slow moving vehicles, bicycles, pedestrians or animals;- have a minimum lane width of 3.5 m.;- have edge line and central markings;- have a head clearance of 4.5 m.;- provide for emergency calling points;- provide for service area’s at a maximum distance of 100 km., directly accessible from the
road, and with 24 hours refuelling possibilities;- have an average daily traffic for single carriageway express road of 5,000 vehicles per
day; for dual carriageway express roads 10,000/15,000 vehicles per day.
Express roads can be designed as single or dual carriageway roads. On dual carriagewayexpress roads intersections can be both grade-separated or at-grade.
Outside urban areas, two road types with a traffic flow function can be identified, namelymotorways and express roads. As indicated by the poor safety records of express roads inChapter 1, motorways are preferred for road-network sections with a traffic flow function.Arguments other than those of safety, however, proved to be decisive factors in all countriesstudied.An analysis of the arguments in the decision to build express roads in several EU-member statesshowed a very consistent image of the arguments and their relative weight. An express road isbuilt or an existing ordinary road is upgraded when the (expected) traffic volume exceeds aparticular number, when higher speeds are considered to be desirable, and when there arefinancial and/or environmental restraints to build a motorway. An economical cost-benefitanalysis is applied in, for example, France and the UK. The argument of better accessibility tomain economical centres alongside the road compared to motorways was always mentioned. Safety as an argument in the decision process was never mentioned, although some designstandards and guidelines do refer to the need for uniformity and continuity in relation to theadjacent network for safety reasons.
In the Netherlands, the main reason for the existence of express roads is considered to be afinancial one. Because of the high costs of motorway construction, cheaper express roads maybe considered as a financially better solution, unless preconditions compellingly force theconstruction of a motorway. The preconditions on which it is decided to choose for amotorway or an expressway are formulated in so called Ministerial Road Plans. In these plansroad sections of the national road network are assigned as motorway or as express road. Theassignment is mainly based on:
- Function of the section within the entire road network: if a section has a less importanttraffic function in the network, the choice for a express road design is more likely.
55
- Traffic volumes: if traffic volumes or estimated traffic volumes exceed standardsdescribed in the national design guidelines, an express road is not a valid option and amotorway design has to be chosen.
Decisions on the Ministerial Road Plans are not made at the level of the Regional Departments,but at the level of the National Ministry of Transport. In the decision making process, allconsequences have to be weighted. In this process the importance of safety (or the difference insafety between motorways and express roads) is often considered to be of less importance thancosts, congestion, accessibility and reliability of the road network. One could say that the actualsafety difference between motorways and express roads is not large enough to make it animportant issue at this decision level.
The analysis of the decisions to build express roads in several EU-member states demonstratethat the choice for the sub-standard express roads will also be made in the future, as a result ofwhich the express road will continue to play an important role in the Trans European RoadNetwork. Replacing existing express road sections by the more expensive motorway sections,or planning for motorways rather than express roads, does not seem to be a realistic strategy.
3.2.2 Improvements in cross-sectional design
Because the express roads will continue to exist in the future road-network, strategies toimprove the poor safety records of express roads have to be developed. An important contribution is the formulation of uniform, safety based design guidelines. Aproblem in this formulation of design guidelines however, is the very small amount of researchinformation on this road type. To overcome this problem in the short term, valuableinformation for the design of express roads can be derived from other, more or less comparableroad types. Though not accurate and specific enough to use as a basis for the formulation ofdesign recommendations or standards, the information can give some insight in possible safetyeffects, probable direction of effects, and estimates on the strength of the effects. Exactquantitative information in this study can therefore not be used directly for express roads.The lack of specific information on safety effects of design parameters on express roadsstresses the need for further research. Information in this study can serve as a guideline in thisfuture research.
Lane widthIn a study of accidents on A-class roads in Great Britain (comparable to express roads asdefined in SAFESTAR), the effects of variations in carriageway width were studied (Hughes etal., 1997):- For dual carriageway sections an increase of one metre in main carriageway width
resulted in a decrease of 56 % in the chance of an accident involving a vehicle joining themain road from an on-ramp.
- For single carriageway A-class roads, within the carriageway width considered in themodel (7.0 - 21.2 m.), a one metre increase in carriageway width at a junction resulted inan estimated accident reduction of 5 %.
- For sections between junctions on single carriageway A-class roads, a one metre increasein carriageway width resulted in a 19 % decrease in accidents (range considered in themodel was 7.1 - 11.5 m.).
56
The authors stress that the observed effects are not linear and that they are only applicablewithin the range of the model.
Based on the analysis of accidents on two-lane rural highways, Zegeer et al. (1981; 1987)concluded that accident rates generally decrease with increasing lane and shoulder width. Laneand shoulder width directly effect run-off-the-road accidents and opposite-direction accidents.Other accident types proved not to be directly effected by these elements. The accident ratesZegeer et al. found were approximately the same for 3.6 m. lanes as for 3.3 m. lanes, possiblyindicating that the limit beyond further increase in lane width are ineffectual. Lane widthsproved to have a greater effect on accident rates than shoulder width.
The evaluation of traffic accidents in Germany by Oellers (1976) led to the result that thefrequency of accidents due to errors in overtaking, being overtaken, and changing lanes washigher on stretches with narrow traffic lanes (3.25 m.).
Hadi et al. (1995) estimated the effects of cross-section design elements on total, fatality, andinjury accident rates for various types of rural and urban highways at different traffic levels.They concluded that significant relationships could be found between lane width and accidentsfor undivided highways and urban freeways. For other highway types, no such relationshipcould be identified. They indicate that for two-lane rural, two-lane urban, four-lane urbanundivided, and urban freeways widening lane width up to 4.0 m., 3.7 m., 4.0 m. and 4.3 m.respectively could be expected to decrease accdient rates. They furthermore indicate that thehighest benefits due to lane widening were estimated for urban freeways, followed by four-laneundivided urban highways, followed by two-lane rural highways. For two-lane urban highwaysthere was significant relationship between pavement width (lane width plus paved shoulderwidth) and accdient frequency rather than between lane width and accdient frequency whencontinuous representations of variables were used. The effect of lane width on accdient rate forthis highway type was lower than other highway types.Accident analysis on express roads in Portugal by Cardoso and Costa (1998) proved that singlecarriageway express roads with lane widths greater than 3.50 m. had better accident recordsthan roads with lower lane widths.
From the above-mentioned results one could get the impression that ‘the wider a road is, thesafer it is’. Michalski (1994) questions the validity of this hypothesis, based on safety researchcarried out in Switzerland. Results showed that increasing the single carriageway width to8.5 - 10.0 m. decreased accident rates as well as the victim rates, but for widths between12.0 - 14.0 m. both rates increased again. For motorways, widening a traffic lane over 3.5 m.causes no significant further improvement of the accident rates. The lane width of 3.5 m. cantherefore be indicated as an optimum for motorways.
Shoulder widthSeveral studies (Hedman, 1990; Zegeer et al., 1988) show a decrease in accidents with anincrease in shoulder width.As noted by Hedman (1990), recent studies show a decrease in accidents with an increase inwidth from 0 to 2 m. Additional benefits for widths above 2.5 m. proved to be very small.Several authors have furthermore concluded that the effect of lane width on accident rates isgreater than the effect of shoulder width.
57
Zegeer et al. (1988) concluded that non-stabilized shoulders, including loose gravel, crushedstone, raw earth, and turf, exhibit greater accident rates than stabilized or paved shoulder. Thesame conclusion was drawn by Armour (1984) who found that the accident rate of roads withunsealed shoulders was between three and four times the accident rate for roads with sealedshoulders. This was true for straight road sections and for road sections with curve or grade.An examination of accident description showed that losing control of vehicle in the gravelshoulder was a contributing cause in about 17 % of fatal accidents.
German comparative studies concerning motorways (Brühning, 1977) show that themotorways with an emergency stopping lane (often 3.0 m.) reduced the total accident rate bymore than 15 %, when compared to rates on motorways with narrow paved shoulders.
MedianA median separates the traffic lanes in opposite direction, thus creating two separatecarriageways. Elements of median design which, according to Zegeer & Council (1992), mayinfluence accident frequency or severity, include median width, median slope, median type(raised or depressed) and presence or absence of a median barrier. Wider medians areconsidered desirable in that they reduce the likelihood of head-on accdientes between vehiclesin opposite directions. Median slope and design can effect rollover accidents and also othersingle-vehicle accdientes (fixed object) and head-on accdientes with opposing traffic. Theinstallation of median barriers typically increases overall accident frequency due to theincreased number of hits to the barrier, but reduces accdient severity, resulting from a reductionor elimination of head-on collisions with opposing traffic.
A 1973 study by Garner and Deen in Kentucky compared the accdient experience of variousmedian widths, median types (raised vs depressed), and slopes on Interstate and turnpike roadsin Kentucky (Garner & Deen, 1973). Highways with at least 9 m. wide medians had loweraccident rates than those with narrower median widths. For wider medians, a significantreduction was also found in the percentage of accidents involving a vehicle crossing themedian.Median slopes of 4:1 or steeper had abnormally high accident rates for various median widths,while a higher accdient severity and higher proportion of vehicle overturn accidents were foundfor medians which were deeply depressed. For median widths of 6 m. to 9 m., the use of araised median barrier resulted in a higher number of accidents involving hitting the median andlosing control. (Unfortunately no information was given on the severity of the accidents).
Climbing laneThe presence of a climbing lane on a two-lane (single carriageway) section can reduce thenumber of catastrophic overtaking accidents that occur due to the presence of opposingvehicles. Such accidents normally involve high-speed head-on or run-off-the-road accidenttypes. On dual carriageway sections the purpose of climbing lanes is mainly to improve trafficoperations. Safety effects are generally caused by the minimizing of speed differences on theadjoining lanes and thereby reducing the chance of rear-end accidents.
Hedman (1990) quotes a Swedish study which concluded that climbing lanes on rural two-laneroads reduced the total accident rate by an average on 25 %, 10 % to 20 % on moderate up-gradients (3 % to 4 %), and 20 % to 40% on steeper gradients. It was also observed that
58
additional accident reduction can be obtained within a distance of about 1 km. beyond theclimbing lane.Martin and Voorhees Associates (1978) found an overall reduction of accidents of 13 % due tothe presence of climbing lanes in the UK.
Harwood et al. (1988) quote a California study by Rinde (1977) at 23 sites in level, rolling, andmountainous terrain where accident rate reductions were found due to the passing laneinstallation of 11 % to 27 %, depending on road width. When the sites in mountainous terrainwere excluded from the analysis, accident reductions of 42 % were found for the level terrainsites as well as for the rolling terrain sites.
3.2.3 Conclusions and recommendations
For both single and dual carriageway express roads, a lane width of 3.5 m. can berecommended.Research on the effects of lane width on motorways and low volume highways resulted in anoptimum lane width of 3.5 m. to 3.6 m. With similar results for road-categories just above andbelow the express road category, it can be safely assumed that this value will also be a safestandard for express roads.
Though the positive effect of lane width is greater than the effect of shoulder width, severalstudies have concluded that the presence of shoulders contribute significantly to traffic safety.The presence of paved shoulders can therefore be recommended for both single and dualcarriageway express roads. Research however does not indicate an unequivocal value for theoptimum shoulder width. Recommendations on the exact shoulder width can therefore not begiven in this report. Local situation and traffic behaviour as well as shoulder widths differwidely in the various European countries.For determining the shoulder width, the method used in the Dutch guidelines(ROA, 1992/1993) seems valuable. Based on the dimensions of the width of a truck of 2.5 m.,a shoulder width of 3.0 m. to 3.5 m. is recommended. A stationary truck on the shoulder can bepassed safely and without influencing traffic operations seriously. Given the lack of empiricaldata and equivocal conclusions, these dimensions can only be seen as ‘best practice’.
The separation of opposing traffic with a median strongly improves traffic safety. The usuallyvery severe head-on accidents are fully excluded.The width of the unpaved median is dependent on the location of the median crash barriers.Accdient barriers are not essential if there is no risk that the median may be crossed. This widthis assumed to be approximately 20 m. Because such a width will generally not be feasible inpractice, medians are normally equipped with accdient barriers. Safety generally improves withwidening of the median. The greater the distance from carriageway edge to median barrier is,the greater the possibility will be that an off-road driver can recover safely (without hitting themedian barrier).Slopes and obstacles in the median should be avoided (or protected).
Because of the considerable improvement of traffic safety, climbing lanes must berecommended on upgrade sections on single carriageway express roads. Even on moderateupgradients (3 % to 4 %), climbing lanes can reduce the accident rates.
59
On dual carriageway express roads, the provision of climbing lanes is dependent on thesteepness of the gradient, the number of heavy trucks, and the possibility of slow moving trucksreducing speeds of other vehicles.
3.3 Rural roads
The effects of different kinds of cross-sectional designs on two-lane rural roads have beenstudied by reviewing existing literature and by analysing Finnish and Danish accident data. Thestudy concentrated on the safety effects (especially injury accidents) of different kinds of crosssections.Experiments were carried out Portugese and Swedish two-lane rural roads with alternativecountermeasures. These experiments have been evaluated.
3.3.1 Accident experience
Although the risk of injury accidents per kilometre driven is higher on urban roads, the risk ofa fatal accident is considerably higher on rural roads. Accident analysis in Finland for instanceshow that about 70% of the injury accidents and 80% of the fatalities occur on two-lane ruralroads. Accident analysis in other European countries show comparable results.The two most dominant accident types of serious accidents are head-on accidents and run-off-the-road accidents. Both accident types indicate the important role of the cross-section designand cross-sectional measures in the safety of rural roads.The accident analysis in Finland and Denmark focussed mainly on head-on accidents and run-off-the-road accidents.The roadway variables possibly associated with these accident types include lane width,shoulder width and type, roadside condition, terrain condition and traffic volume. Also, thepresence of unprotected road users can have an effect on cross-section safety.
On two-lane 80 km/h rural road stretches (also 100 km/h in Finland), 50% of the injuryaccidents in Finland were run-off-the-road accidents and 23% were head-on accidents. InDenmark, the percentages were 31% and 16% respectively.In fatalities, head-on accidents covered 55% in Finland and 39% in Denmark. The percentagesfor run-off-the-road fatal accidents were 20% in both countries.
Based on the data of the Accident Investigation Teams in Finland, it was possible to find somedifferences in the causes for head-on accidents and run-off-the-road accidents. Head-onaccidents had more often several explaining factors leading to an accident. As for the run-off-the-road accidents, it was often possible to find one strong factor affecting the occurrence of anaccident (e.g. paroxysm, alcohol, falling asleep). In head-on accidents, bad weather and badroad conditions easily led to driver errors.If there is one major factor leading to the occurrence of an accident, the possiblecountermeasures are somewhat limited. Most of these factors cannot be eliminated easily (e.g.paroxysm, alcohol).
60
Run-off-the-road accidentsAccident analysis in Denmark proved that speed, wheels on the verge or soft shoulder, objectson or along the road and alcohol were frequent causes of accidents and determining factors ofthe severity of the accidents.High speeds (in relation to the road alignment or to weather conditions) were described as acause in at least 23% of the run-off-the-road accidents. Entering the verge with the right handwheels of the car was specifically mentioned in 19% of the accidents.
Study of the data of the Accident Investigation Teams, for all victims in fatal run-off-the-roadaccidents on two-lane rural road stretches showed the following results:- Run-off-the-road accidents accounted for only 23% of the number of victims.- Almost 60% of the victims resulted from run-off-the-road accidents on straight or almost
straight road sections.- Relatively large numbers of run-off-the-road victims occurred on straight or almost
straight road sections, where the driver had fallen asleep.
According to earlier in-depth studies in Finland (e.g. Räsänen, 1997), almost 80% of fatal run-off-the-road accidents could be attributed to a single explanatory variable. According to thisFinnish study, the drivers’contribution to these accidents can roughly be divided into fourcategories: alcohol, driving abilities (vehicle handling, perception, prediction), driverinattention, and suicide.
Figure 3.1 Major causes in run-off-the-road accidents
Head-on accidents:Study of accident causes of fatal head-on accidents on Danish two-lane rural roads, showedthat causes were mentioned in about 50% of the accidents. Causes which were often mentioned were high speeds, overtaking manoeuvres, and wheels on the verge or soft shoulder (resultingin overcorrecting manoeuvres). Driving at a speed too high for the road alignment or weatherand road conditions, and thereby losing control over the vehicle, was stated in many of theaccidents.
A recently published Danish report (AVU, 1997) showed that the most frequently occurringaccident factors were high speed and driving under the influence of alcohol, narcotics and/or
61
medicine. Other mentioned causes were: errors in connection with overtaking and recklessdriving, improper evasive actions, inattention towards the road (e.g. using mobile telephones),and tiredness.
Study of the Accident Investigation Teams data for all victims in fatal head-on accidents ontwo-lane rural road stretches showed the following results:- Head-on accidents accounted for 77% of the total number of victims.- Overtaking proved to be a quite rare event in head-on accidents (10% of the victims),
whereas one would expect this to be a more important accident cause.- Almost 90% of the victims in head-on accidents resulted from an accident occurring on
straight or almost straight road sections.- Head-on accidents were more often connected to bad weather conditions, especially
slippery roads, compared to run-off-the-road and other accidents.
Figure 3.2 Major causes in head-on accidents
Two major things that distinguish head-on accidents from run-off-the-road accidents were:- Head-on accidents had more often several explaining factors leading to an accident.
Weather and road conditions played a more important role in head-on accidents.- In contradiction with run-off-the-road accidents, alcohol was not involved in head-on
accidents more than in accidents on average.
3.3.2 Effects of cross-sectional design
Two-lane rural roads include many different types of road, ranging from traditional, windingroads to modern high quality roads with gentle curves and full cross-sections. Thus, undividedrural roads can have considerable variation in dimensions of traffic lanes and shoulders, not tomention the variety in roadside characteristics.Of the cross-section design parameters, only the concept of lane width is fairly unambiguousbetween different countries. Other design parameters, such as shoulder width and type, orsideslope design, vary widely in design practice.
In the accident analysis in Denmark, accident risks were calculated for rural roads of differentroad and shoulder widths. The main conclusions of this study were:
62
- The average risk for run-off-the-road accidents was somewhat lower on wider roads.- The average risk for head-on accidents was not found to be influenced by the road width.- On 6 m. and 7 m. roads, the average risk for run-off-the-road accidents was lower on
wider paved shoulders.- The average risk for head-on accidents was not found to be influenced by the paved
shoulder width.- For all other accidents, the average risk in general was lower on wider roads and roads
with wider shoulders.- A previous Danish study (Danish Road Directorate, 1996) showed that about 70% of all
obstacles in the accident were situated at a distance of 3 m. or less from the road, andthat most of the accidents might have been avoided if the distance between the edge lineand the obstacle was 9 m. or more.
- In fatal accidents in Finland, speeds were often high, judging from the fact that vehicles inmore than 50% of the accidents were extensively or totally destroyed. Almost all of therun-off-the-road accidents took place quite close, inside 7 m, from the road edge.
According to these Danish results, it can be concluded that in general the upgrading of the roaddoes not seem to effect the average risk for head-on accidents.These results are however in contrast with results of studies in the USA, Germany andAustralia, where widening of both lanes and shoulders showed a decrease in head-on accidents.
Driving laneStudies in the USA and Germany (Zegeer, 1987; Zegeer, 1995; Brannolte et al. 1993) provethat the number of accidents generally decreases with increasing lane width. Accident rates didnot decrease any further for lane widths of 3.3 - 3.6 m., possibly indicating the lane widthbeyond which further widening were ineffectual. Lane widths of 3.3 - 3.6 m. (Brannolte et al.mention 3.5 m.) can be seen as an optimum.
Zegeer et al. (1987) quantified the effects of lane width on highway accident experience, basedon an analysis of data for nearly 8000 km. of two-lane highway from seven states. Resultsshowed that the accident types found to be most related to cross-section features include run-off-road, head-on and sideswipe accidents. Lane widening of 0.3 m. was found to reduce theserelated accidents by 12%, 0.6 m. by 23%, 0.9 m. by 32% and 1.2 m. by 40%.The results conclude to be valid for two-lane rural roads with lane widths of 2.4 to 3.7 m. andADT’s of 100 to 10,000. The study was reported to control for many roadway and trafficfeatures, including roadside hazard, terrain and average daily traffic. However, to fullyeliminate the effects of all other variables is impossible, and therefore the effects of lanewidening may be overestimated.
ShoulderIn general there is no agreement in EU countries on the paved shoulder width of non-motorways.Situations vary from non-motorways without shoulders to non-motorways with wide shoulders.As a result, pavement width can vary from 6 m. or less, to 11.5 m.These variations partly reflect the fact that there is no consensus on the safety effects of theshoulders.
63
A Danish study (Vejdirectoratet, 1984) concluded that for carriageway widths of 6.5 to 8 m., apaved shoulder increase from 0.2 m. to 0.5 m. showed a significant reduction of accident risksfor vehicle accidents by about 25% and for pedestrian and cycling accidents by 40%. Theeffects of further increase in shoulder width (� 0.9 m.) was uncertain, but indicated no effect onvehicle accidents and possibly a further reduction of 20% in pedestrian and cycling accidents.
In Germany, Brannolte et al. (1992) detected that two-lane roads with shoulders had anapproximately 10% lower accident rate compared to similar roads without shoulders. Thedifference in accident cost rate was greatest when turning and crossing accidents were includedin the comparison.
In the USA, Foody and Long (1974) found that the mean accident rate for stabilised shouldersections was significantly lower than for sections of unstabilised shoulders. Particularly, theresults indicated that shoulder stabilisation or paving was quite effective in reducing run-off-the-road accidents on narrow carriageways, typically 6 m. or less in width, but had only littleeffects on road widths of 7.2 m. or more.
Several authors (Zegeer, 1979/1987/1995, Brannolte et al. 1993, Edholm & Roosmark, 1969)found decreasing accident rates with increasing paved shoulder widths. An optimum value forthe shoulder widths is not determined unambiguously, but a total pavement width of 10 m. (3.5- 3.7 m. wide lanes and paved or stabilised shoulders of 1.3 - 1.5 m.) is mentioned as thepavement width beyond which further widening does not improve safety. Pavement widthsgreater than 10 m. could possibly encourage higher speeds and more careless driving.In general, ‘overwide’ roads have proven to cause increasing accident rates.
Pavement widthIn Sweden, the effects of pavement width on the accident rates have been modelled from the1970's (Brüde & Nilson, 1976, Brüde & Larsson, 1977). The model differentiates betweenthree classes of alignment, namely:- mainly straight sections with slight gradients;- comparatively large curve radii and slight or steep gradients;- small curve radii and slight or steep gradients.For all 90 km/h roads the accident rate decreased with pavement widening up to about 10 m.For further increase of the pavement width , the accident rates increased in the third alignmentclass (small radii and slight or steep gradients).
In Finland, Peltola (1995) concluded that on main roads, the risks for personal injury andfatality appeared to be lower on wide pavements (medium wide: 8.1-9 m.; wide: over 9 m.).The results were equal for ADT’s over and under 6000 vehicles per day. However, on otherthan main roads, the fatality risk seemed higher on wider pavements. Increase in risk wasattributed to more severe single-vehicle, overtaking and head-on accidents.
Carlsson and Lundkvist (1992) and Brüde and Larsson (1994) studied the safety effects of anincrease in lane width (from 3.75 m. to 5.50 m.) and a simultaneous reduction of shoulderwidth (form 2.75 m. to 1.00 m.). The results showed that lateral positioning was changed sothat passenger cars were positioned further to the right of the roadway. Furthermore, the
64
variance of the lateral position was increased. On typical 13 m. wide roads, the severity ofaccidents increased in connection with the widening of the driving lane, but on express roadsthe cross-sectional change reduced the accident risk. Due to small accident counts clearconclusions were not drawn.Later, in 1996 Brüde and Larsson compared existing roads with wide lanes and narrowshoulders with remaining roads of the same type but without wide lanes (having wide shouldersinstead). The accident rate, injury consequence and injury rate were almost the same for roadswith and without wide lanes.However, there were differences in accident types. Wide lanes had a larger proportion of singlevehicle accidents, but instead fewer rear-end, turning and crossroad accidents. It was concludedthat wide lanes are not a practicable way of finding less expensive traffic safety alternatives tomotorways.
In Germany, Brannolte (1993) also studied the effects of overwide rural roads with 5.25 m.lane width. Compared to a typical 3.75 m (or 3.5 m.) lane width, and 2 m. (1.5 m.) unpavedshoulder width, the accident rates and accident cost rates were higher on overwide roads.
Roadside designAccording to Ruyters et al. (1994) in most European countries approximately one quarter of allcasualties is killed in accidents with obstacles. Obstacles by the roadside (e.g. columns andtrees) affect the accident risks and especially the accident severity. To reduce the number ofcollisions with obstacles, obstacle free zones are recommended. Based on accident research,widths of the obstacle free zone from 7 m. to 15 m. are recommended.
A Danish study (Danish Road Directorate, 1996) showed that about 70% of all obstacles in theaccident were situated at a distance of 3 m. or less from the road, and that most of theaccidents might have been avoided if the distance between the edge line and the obstacle was 9m. or more.In fatal run-off-the-road accidents in Finland, almost all of the accidents took place quite close,inside 7 m, from the road edge.
Zegeer (1988 and 1992) proved that on slopes steeper than 5:1, the risk of rollover accidentsincreases. Because rollover accidents are known to be severe, sideslopes steeper than 5:1 arestrongly advised against.
RecommendationsAlthough not all results and recommendations are completely unequivocal, the followingdimensions for cross-section design can be recommended:Lane widths on two-lane rural roads should preferably be about 3.5 m.The presence of shoulders is recommeded because of the positive safety effects. Therecommended width of the shoulder is between 1,3 m. to 1,5 m., resulting in a total pavementwidth of approximately 10 m.Sideslopes steeper than 5:1 are not recommended, because of a significant higher severity ofaccidents with vehicles that run off the road.Recommended values for obstacle free recovery zones differ widely in the studies used, andvary between 3 m. and 15 m.
65
3.3.3. Effects of alternative countermeasures
If accidents are caused by many factors (e.g. head-on), affecting one factor could have potentialin preventing the occurrence of an accident. However, as the factors are many and accidents donot seem to accumulate into black spots, preventive measures should be very extensive orgeneral, and probably also expensive. Costlier solutions can be justified on high volume mainroads where most head-on fatalities occur, as the expected accident reduction can be predictedto be quite significant.
If an accident has one strong explanatory factor (e.g. run-off-the-road), possiblecountermeasures against the cause of the accident are somewhat limited. Many such factorscannot be eliminated by road design (e.g. seizure, stroke). However, there are potentialmeasures for e.g. increasing driver alertness. The consequences of these accidents can yet beminimised.
In general, means of affecting severe accidents fall into two categories: either preventing thecause (i.e. preventing one or several contributing factors to an accident) or reducing theseverity of an accident where conditions for an accident exist.It appears that road design measures available for preventing the causes of severe accidents arefairly limited, and new ideas should be encouraged. Therefor every potential measure should beconsidered carefully. Furthermore, the fact that one might affect only the consequences shouldbe accepted.
It is estimated that the median barrier eliminated or greatly reduces the fatal outcome of head-on accidents on two-lane roads. However it will possibly create a new hazard on the road. Bynow (ETSC, 1998), it has been estimated that median barriers will increase accident rates by30% even though they reduce accident severity.
Several case studies on the effects of different types of median separators were carried out inthe Safestar project. These case studies were conducted in Sweden and Portugal.Behavioural studies were carried out with an instrumented car to evaluate the safety potentialof the separators, with an unobtrusive microscopic observation of the test drivers.Because of the short period since the implementation of the test sections, accident data are notavailable yet.The tests in Portugal and Sweden will be discussed separately.
PortugalDifferent types of separation of different directions of traffic on narrow four-lane roads weretested in the south of Portugal (Cardoso, 1998), with the use of an instrumented vehicle.Tested were a concrete barrier, plastic delineators and a dual centre line.The test route was 13.5 kilometres long. The selected four lane road was 14 m. wide (3 m.wide driving lanes and 1 m. wide shoulders).Special attention was given to the driving behaviour in the left lane. Driving speeds and lateralposition of the test runs were analysed.
66
The distance to the left edge line proved to be significantly greater for the more solid types ofmedian separations. This distance to the left edge line was 0.6 m. for a dual centre line, 0.7 m.for plastic delineators and 0.8 m. for a concrete barrierOn the concrete separator sections the mean speed was 92 km/h, on the sections with a doubleline the mean speed was 80 km/h, and on the sections with plastic delineators the mean speedwas 78 km/h. The speed differences were not statistically significant, and there proved to be norelation between speed and distance to the left edge line.
Costs per kilometre were 35,000 ECU for plastic delineators, 2,000 ECU for a dual centre line,and 36,000 ECU for a concrete barrier.
SwedenIn a Swedish project measurements were carried out on an experimental road in Gävle-Axmartavlan. test drivers drove along a test section with six different cross sections in aninstrumented vehicle.The following cross section types were studied:• Two lane road with wide shoulders (total width 13 m.); speed limit 90 km/h.• 2+1 road without median barrier; speed limit 90 km/h.• 2+1 road with median barrier; speed limit 90 km/h.• Wide two lane road (5.5 m. driving lanes and 1 m. shoulders); speed limit 110 km/h.• Typical motorway.• Narrow motorway (23 m.).
Speed measurements gave the following results:
Road type Standard deviation for speed ofindividual drivers drivers
(kn/h)
Standard deviation formean speed
(km/h)
Mean speed(km/h)
Typical two-lane road (13 m) 10,6 13.0 89.2
2+1 without median barrier 12.2 14.9 91.9
2+1 with median barrier 12.0 15.8 91.3
Wide-lane road (13 m) 12.6 15.3 97.8
Typical motorway 8.8 12.6 105.0
Narrow motorway 10.6 12.2 102.1
91.9: average for two-lane road sections
Table 3.1 Mean speeds and standard deviations for the speeds on different types of roads
After driving the drivers were asked to fill in a questionaire. The answers were given on a five-point scale (1=bad, 2= not so good, 3=good, 4=very good).To the question “How did you experience the road design?”, drivers gave 2+1 roads with andwithout a barrier as well as typical two-lane road sections an average rating of 2.5.Narrow motorways were given a rating of 3.9, and wide two-lane road section were give arating of 2,6.
To the question “How did you experience overtakings?” drivers gave an average rating of 2.9for 2+1 roads both with and without a barrier. Two-lane roads got a slightly lower rating of
67
2.4. Wide two-lane roads were rated 3.0. Narrow motorways were experienced to be better(4.0) compared to other road types in the study.
Total costs for a 14 km Gävle-Axmartavlan experiment road (2+1 road with cable wire) were3.5 million ECU which is roughly about 250 000 ECU per road kilometre. Individual expensesper road kilometre (or per unit) were as follows:• cable wire 27 000 ECU,• road markings together with traffic signs 19 000 ECU,• widening stabilised shoulders together with 1:6 sideslope 85 000 ECU,• pavement 66 000 ECU,• parking areas 108 000 ECU each (total 2).
68
4. SAFETY DEVICES
4.1 Motorways and express roads: criteria for safe roadsides in relation to the installa-tion of safety barriers (steel and concrete)
4.1.1 Introduction
To protect occupants of vehicles that leave the road from serious injuries, safe roadsides andmedians are important. Free-zones, safety barriers, and impact attenuators are effective torealising this.The CEN standards for safety devices, that are currently being developed, ensure theeffectiveness of these devices, but say nothing about the road characteristics and circumstancesin which they should (or should not) be applied. This research aims to define criteria for places where safety devices are necessary. This is basedon a general design philosophy for safe shoulders on motorways (and express roads), and basedon design criteria for safety devices. There is also a need for criteria to chose between steel and concrete barriers; the containmentlevel of barriers will be a part of these criteria.
The target groups of these criteria are the road authorities in the TERN-framework, nationalroad authorities in the European countries, authorities in transport departments, and (technical)staff responsible for road design and/or safety devices. More uniformity concerning safeshoulders on European roads will be the final goal.
Methods usedThe research was carried out by means of a literature study included the national Europeanstandards. Furthermore, a questionnaire was prepared and sent to European institutes andministries. This questionnaire contained, for instance, questions about national standards and/orcriteria for the use of safety barriers, and accident data on motorways and express roads wheresafety barriers were involved. There was also a request to send copies of recent researchreports concerning these subjects and in particular about cost-benefits. The European accidentswere completed with data from an accident study carried out by SWOV. Data from European countries, but also from the United States were analyzed for preparing aproposal for standards and strategies for EU-countries.
QuestionnaireThe questionnaireA questionnaire for safety barriers on motorways and express roads wasmade and sent to all specialist of European countries.The following subjects were asked about:* national standards and/or criteria for making the decision to locate safety barriers * the width of the obstacle-free zone, if there is no need for a safety barrier;* containment levels of the national barrier construction types;* presence of safety barriers with a distinction in steel and concrete (rough estimation);* accidents on motorways and express roads where safety barriers and off-the-road
accidents were involved; accident data were asked with the following characteristics:number of injury accidents and number of fatalities and injured persons;
69
* copies of recent research reports concerning safety barriers and particularly thedifferences between steel en concrete barriers, together with aspects such as costs,accidents, and cost-benefits.
Questionnaires were sent to 16 European traffic safety institutes or ministries; 13questionnaires were completed, a response of approx. 80%. In some cases when a country didnot give any data on a subject, we filled in the missing value based on the literature resource.It seems that most of the countries have standards both for motorways and express roads.
From data on injury accidents with safety barriers on motorways from some Europeancountries, it seems, however, that erecting safety barriers is not always an absolute safesolution. Accident data showed that approximately 20% of the injury accidents on motorwayswas the result of a collision with a safety barrier.
4.1.2 Roadside
Attention has to be paid to the reduction of the high percentage of injury accidents withcollisions with safety barriers and obstacles. The following line of action for creating a conceptof a safe roadside is general accepted. The list is similar to the strategy used in the UnitedStates called ‘create forgiving roadside’. 1. a shoulder without obstacles (and without safety barriers);2. a shoulder with safe slopes;3. a shoulder with fixed objects that yield easily upon collision;4. a shoulder with accdient cushions;5. a shoulder with an effectively functioning safety barrier.
From the list of five possible solutions, the first four can be qualified as the best. The next bestis the erection of safety barriers; this subject will be extensively discussed.Owing to the systematical research carried out in Europe concerning this list of five possiblesolutions, much research is quoted from investigations carried out by SWOV under theauthority of the Ministry of Transport and Public Works. In addition, research and data fromstandards from other European countries is mentioned.
A shoulder without obstaclesThe question that immediately rises when discussing an obstacle-free-zone is how wide thiszone should be. Every report beginning with this topic refers to American research from the1960's and 70's. Since that time, as far as we know, hardly any more studies on this subjecthave been carried out in the United States. Although these studies were extremely valuable andhave been used as a guiding principle in many European countries, their figures are based onthe American situation. Two factors in these studies which differ considerably from the currentEuropean situation are the differences of vehicle mass and driving speeds.
The only known study carried out in Europe into a desirable width for an obstacle-free-zonewas done in the Netherlands in the 1980's. This study involved road sections lined with rows oftrees; these rows being located at various distances from the edge of the road. What thisresearch establishes is the relationship between the accident ratio and the distance that vehicles
70
travel onto the shoulder when an accident occurs. This ratio is the number of accidentsinvolving trees as opposed to the number of accidents nót involving trees.
For motorways in the Dutch study it was found that when trees are planted at a distance ofapproximately 10 metres from the road, 10 out of the 100 accidents occurred with trees(figures based on a significant regression curve). The distance is measured from the markingline of the right-hand side traffic lane.
For single-lane highways (in some cases also express roads) 10 out of the 100 accidentsoccurred with trees planted at a distance of 7 metres from the road. The distance is measuredfrom the border line on the right traffic lane. For the Dutch Ministry of Transport and PublicWorks these values were used for preparing standards.When these distances of obstacle free zones were compared with the values of the standards ofother European countries, agreement was found. The next table gives these values based on thementioned questionnaires.
Germany 6 (10 if dangerous zone) 4.5 (7.5 if dangerous zone)
Greece 9 (19 near railway roads) 9 (19 near railway roads)
Finland 7 5.5 - 6.5
France 10 8.5
Netherlands 10 (if v=120 km/h: 13 m) 6
Norway 6 (if ADT �15,000) 5 (if ADT is high)
Portugal 3.5 3.5
Sweden 10 (if v =110 km/h) 9 (if v = 90 km/h) 7 (if v = 70 km/h)
10 (if v =110 km/h) 9 (if v = 90 km/h) 7 (if v = 70 km/h)
Switzerland 12.5 5
United Kingdom 4.5 4.5
1) In Denmark, the width is in discussion as a result of an audit concerning the design of the roadside as example. Theintention is that the process will be based on effectiveness studies.
Table 4.1 The width of the obstacle free-zone based on the following question on theSAFESTAR questionnaire ‘What is the width of the obstacle free-zone if there isno need for a safety barrier?’
71
A shoulder with safe slopes
United StatesIn the United States graphs have been developed with as basis that slopes with an angle of 4:1and flatter are recoverable. Vehicles on recoverable slopes can usually be stopped or steeredback to the roadway. Slopes steeper than 3:1 are critical and are usually defined as a slope onwhich a vehicle is likely to overturn.
The NetherlandsThe only study for slopes ever carried out in Europe has been an investigation by SWOV.Mathematical simulations formed the basis for this research. The simulation results have beenverified by using twelve full-scale tests on slopes with gradients of 2:1 and 4:1.From this study it was found that the radius of curvature at the top of the slope was of greatimportance in preventing the wheels from leaving the ground. For declining slopes, therefore,the radius of curvature may not be any smaller than 9 metres, but should preferably be 12metres. With a gradient of 4:1, the vehicle stays in good contact with the ground, but steeringmanoeuvres are not helpful in gaining control. If the driver wants to be able to get the vehicleon the slope under control, a gradient of at least 5:1 is necessary for high slopes (e.g., 5metres). For lower slopes (approx. 2 metres), a gradient of at least 6:1 is required.Ascending slopes were also studied by SWOV by using simulations of braking and steeringmanoeuvres. It was found that the radius of curvature at the foot had to be at least 4 metres,and that a gradient of 2:1 or gentler would be acceptable.
United KingdomIn United Kingdom safety fences should be installed at trunk roads where speeds of 50 mph orabove are allowed, in the following situations:- on the top of an embankment with a height of 6 m or more;- on other embankments where there is a road, railway, water hazard and others features at ornear the foot of the slope;- on the outside of curves less than 850 m radius on embankments between 3 and 6 m in height.
FranceOn motorways safety fences are prescribed if the height of the top is more than 4 m and 1 m ifthe area at foot level is dangerous with a length of at least 30 m. Fences are not necessary if theslope is ‘soft’, i.e. an angle of 1 : 4 or more. Motorways in South France must be provided witha safety fence if the slope height is between 2.5 and 4 m.
Germany For the German motorways a division is made into the longitudinal road radius and the distanceof the slope to the edge. If the outside radius is more than 1,500 m, safety barriers are requiredif:* a slope of < 8:1 at a distance of less than 6 m (10 m)* a slope of 8:1 to 5:1 at a distance of less than 8 m (12 m)* a slope of > 5:1 at a distance of less than 10 m (14 m).Numbers between brackets means that, in the case of a very dangerous zone at the foot of theslope (for example deep water), the distance has to be increased.
72
If the outside radius is less than 1,500 m, add 4 m to the given distances (and 2 m for thenumbers between the brackets).For the German undivided roads with an outside radius of more than 500 m, the dimensions, ifsafety barriers are required are:* a slope of < 8:1 at a distance of 4.5 m (7.5 m)* a slope of 8:1 to 5:1 at a distance of 6 m (9 m)* a slope of > 5:1 at a distance of 8 m (12 m).If the outside radius is less than 500 m, add 6 m at the given distances (and 4 - 5 m for thenumbers between the brackets).
Switzerland A graph is given of the relation between slope height and the necessity to install a safety fence.For motorways: it ranges from a flat shoulder with a obstacle free-zone of 12.5 m to a slopewith a height of 10 m with an obstacle free-zone of 27.5 m.For undivided roads the range is: from a flat shoulder with a obstacle free-zone of 5 m to aslope height of 7 m with an obstacle free zone of 20 m. A slope angle is not given.
DenmarkThe criteria in Denmark are based on the Dutch mathematical study.
Sweden In Sweden a slope angle of 5:1 for downwards slopes is preferred.
Shoulder with fixed objects that yield easily upon collisionIf a fixed object is made to yield, it can be placed in an obstacle-free zone without safetybarriers. To reduce the impact severity for cars an appropriate breakaway device can be used.Breakaway supports refers to all type of sign, luminaire, and traffic signal supports that aredesigned to yield when hit by a vehicle. The release mechanism may be a slip base, plastichinges, fracture elements, or a combination of these. In the United States criteria to determine if a support is considered as ‘breakaway’ aredescribed in ‘Standard Specifications for Structural Support for Highway Signs, Luminairesand Traffic Signals’. The CEN is preparing standards for testing fixed objects (Passive safety ofsupport structures for road equipment).
Examples of collision-safe fixed objects for the European situation are:- aluminum lighting poles with a length of 10 metres and smaller, and steel poles with a
slipbase; a deformable (patented) steel lighting pole developed in Sweden in the 1970’s.NB. In Sweden roadside safety experts are trying to change the policy to locate lightingpoles on the outside of curves onto insides of curves.
- a telephone box on a thin pole that bends forward and does not break off during acollision, thus preventing the pole from flying through the windscreen.
- signs on thin poles that easily bend during a collision; larger direction signs on thin polesin an A-shape.
- drainage features such as culverts and ditches have to be constructed with flattened sidesin such a way that these constructions are traversable.
73
Although fixed objects probably present more of a danger for riders of motorcycles than formotorists in the case of an off-the-road accident, a shoulder with solitary obstacles is much tobe preferred, in terms of motorcyclist safety, above a shoulder that is completely shielded by asafety barrier.
Shoulder with crash cushionsIf solitary rigid obstacles along a shoulder cannot be removed, they can be shielded with a crashcushion. Crash cushions are applied on motorways in mainly two different situations: in pointedareas at exits (often at the beginning of a safety barrier) and on shoulders to shield single ob-jects. If crash cushions have been hit head-on, the vehicle usually remains within the shoulderso that it forms no danger for other traffic. In the case of a side impact, most types of crashcushions function like a safety barrier.
Different European countries have their own type of crash cushion (Italy, United Kingdom,Germany and the Netherlands). Most accident experience has been gained in the Netherlandsbecause the first crash cushions were located in 1982. In 1989 an evaluation study was carriedout at a moment that at 170 locations, crash cushions were installed. The study dealt with 97collisions with the crash cushion. Only 6 collisions resulted in (slight) injuries. At this momentmore than 350 crash cushions are installed in the Netherlands.
Shoulder with safety barriersIn the concept of a safe roadside, protecting the roadside (shoulders and median) with safetybarriers is the least safe solution. An effectively functioning safety barrier prevents a vehiclefrom leaving the roadway and striking a fixed object or terrain feature that is considered morehazardous than the barrier itself. But as already shown, a collision with a safety barrier is neverfree from the risk of injuries for the occupants of the colliding vehicle, as well as for other roadusers.
The requirements applying to safety barriers are:1. The effective guiding of vehicles that have run off the carriageway.2. This guiding function must remain after the collision. In general, it can be said that if the
first requirement is satisfied, the second one will be also.
The effectiveness of the guiding can be further qualified by the following criteria:- Roll angle must be kept to a minimum.- Occupants must not suffer any serious injury.- The exit angle must be small (to avoid accidents with third parties).- Specifically for medians and verges between the roadway and the cycle track/footpath:
the construction and the vehicle (or parts of them) may not wind up on the other side ofthe road, putting them in the way of oncoming traffic.
CEN standardsThese assessment criteria have been described quantitatively in terms of standards for testingsafety barriers (CEN/TC 226, prEN 1317). These CEN tests give a good picture of the degreeof safety provided by the tested safety barriers under test conditions. Both flexible steelconstructions and rigid concrete constructions appear to satisfy the standards. In this sense, the
74
tests are valuable for distinguishing good constructions from bad ones and for enabling thecomparison of one kind of construction with another. The CEN tests, however, are based on ‘straight’ input conditions. Accident under conditionssuch as slipping, braking, and steering manoeuvres are not involved; it is also hardly possible torealize the many conceivable accidents. Mathematical simulations offer more possibilities in thisregard; verification tests must always be a part of these simulations.
Containment levelsThe levels of vehicle containment within the CEN-standards are linked with the severity ofimpact tests that safety barriers should undergo. The assessment of the performance of thebarrier is based on criteria of the impact severity and the working width (deflection). If the testis successful, the barrier is ranked within a class of containment level. These levels are dividedin low angle containment (T1-T2), normal containment (N1 and N2), higher containment (H1 -H3), and very high containment (H4a and H4b).The results of investigation by means of the questionnaires give us the information about thequalifications from the European countries of the safety barriers in their own country. Most ofthe countries qualify their safety barriers in standard situations as ‘normal containment level’(N1 -N2) for steel barriers, and as ‘higher level’ (H1 - H3) for concrete. You wonder howmany countries have CEN-test results for these barriers.
The next question is very important: on which type of road had to be installed which type ofbarrier? Here also the questionnaires give some insight. Asked for were the criteria to place ahigh performance barrier. The most frequent answers given were:- danger for lower sited roads, buildings;- others (bridge parapet, noise protection, water areas, rail crossing);- narrow median;- percentage heavy traffic.
Some European countries have made a beginning with these criteria in their standards. Switzerland’s standards regulate the application of level ‘H2' (as the highest class) in the caseof the protection of hazards with large accident risk. For shielding of railroads and chemicalindustry plants, a H2-level is also recommendedIn Germany a concrete barrier is recommended if the risk for collapse of the barrier is too high.As criterium for a high traffic flow is mentioned 50.000 vehicles in 24 h. Drafts are preparedwith characteristics of the following items: accident history, traffic volume, percentage of heavytruck traffic, number and width of lanes, and radius of curves.
Literature study into safety barriers at H4 levelSWOV carried out a literature study on safety barriers at a very high containment level. Testsdone in Europe were according to the H4 level of the prEN 1317 standard; it was establishedthat not many full-scale tests at this level had yet been carried out up till now. In addition, other tests on a similar level as H4 tests are described. These tests were carried outin Japan and the United States and deviate from the H4 tests in that they involve vehicles with adifferent mass and a somewhat different collision speed and/or collision angle. For inclusion,however, the collision energy was of the same level.From the research, the following conclusions were drawn:
75
- Heavy vehicle safety barriers can be made of either steel or concrete. Examples ofconstructions made of both these materials were found that satisfy the desired H4 level.
- For constructions with small widths, concrete is to be preferred over steel constructions;and for constructions with greater widths, steel is to be preferred.
- The available heavy vehicle safety barriers are higher than current constructions. Vehiclesafety barriers with a height of about 1.3 metres appear to provide good results. With aheight of about 1.0 metre, vehicle roll-overs (overturning) still occur. Constructions thatare 1.3 metres and taller have a positive effect on arresting cargoes.
- The damage suffered from collisions involving a steel construction appears to be muchgreater than damage suffered from collisions involving concrete safety barriers.
- It appears possible that the ASI (Acceleration Severity Index) values for passenger carsduring a collision with a heavy vehicle safety barrier, are below the highest permittedvalue of 1.4 in the CEN standard.
Experiences with mathematical simulationsMathematical simulations are very helpful to confirm the effect of construction modifications.Some examples can be shown here.The first example is a study of the effect of the degree of flexibility of a steel safety barrier onvehicle’s decelerations and exit angles. SWOV found that the exit angle at a collision against aflexible construction is an average of 5� smaller than at a collision against a less flexibleconstruction.The second study of construction modification involved the coefficient of friction of the surfaceof the concrete New Jersey barrier. Established is the effect of this friction on the climbingheight of the vehicle upon collision. It was found that a reduction in the coefficient of friction of50%, reduces the climbing height up to c. 20 cm, and so the risk of overturning.
During the last years, SWOV also carried out research in the field of movable barriers. Withcomputer simulations, for example, the strength of the connections between the blocks(concrete or steel) and the movement in lateral direction under impact has been calculated.
4.1.3 New developments
Steeper profiles of concrete barriersThe fact that the smaller passenger cars have a greater risk for overturning has led a number ofEuropean governments (the UK, the Netherlands) to abandon the New Jersey profile and tostart using a steeper profile. Although the vehicle's rate of deceleration has somewhatincreased, the number of cars expected to overturn is fewer.Since a steep profile easily leads to damage to the body of the vehicle, the latest development inthe Netherlands is the 'Step barrier'. This is a barrier with a steep profile accompanied by asmall upright edging at he bottom. Simulations carried out by SWOV show that this edgingdoes not unfavourably affect the course of a collision.
The same results with steeper profiles were established earlier in the United States. Tests withcars with a weight of 815 kg on a vertical-faced concrete wall have shown that such a barrierminimizes vehicle rotation on the longitudinal axis. The vehicle deceleration levels are greaterthan for concrete barriers with a specially front shape, and the exit trajectories are with a higherarc going away from the barrier.
76
Need for modifying barriers Changes in the cross-sections of motorways also makes it necessary to modify safety barriers.Examples of these changes are:- more traffic lanes for each carriageway which can result in larger crash angles;- narrow medians that necessitate the use of narrow safety barriers;- due to increasing traffic concentration, there is a greater need for safety barriers that are
maintenance-free and are not seriously damaged during a collision;- some countries have separate lanes for heavy traffic; a physical separation of truck traffic
from other traffic with a barrier is desirable. In these cases there is a need for barrierswith different collision properties on either side.
- the increase in heavy truck traffic and buses with a high centre of gravity is necessitatingthe use of high containment constructions; there are developments in the UnitedKingdom, Switzerland, Italy, and the USA.
Constructions for single-lane roadsAlthough construction modifications have the potential for favourably affecting the outcomesof accidents involving safety barriers, vehicular manoeuvres made before the accident, as wellas the driver's influence on the path of the vehicle after the collision, are more important. Incooperation with industry, SWOV is now developing a safety barrier for single-lane roads thatshould allow the vehicle to remain close to the construction and thus avoid the danger ofsecondary accidents. Initial full-scale tests with a collision speed of 50 km/h produced goodresults.
The difference between steel and concrete barriersBased on the experiences with full scale tests and mathematical simulations, the following isconcluded for steel and concrete barriers:- the severity of the accident, in terms of vehicle deceleration, is greater when a concrete
barrier is hit;- when a vehicle hits a concrete barrier with a special profile (no steep profile), the car’s
front end leaves the ground; especially in the case of smaller passenger cars, there is therisk of overturning;
- the exit angles are larger for concrete barriers.
Based on the investigation results by means of the questionnaires, it can be concluded that roadauthorities already make a distinction between steel and concrete barriers, based on dailypractice. In the questionnaires it was asked for the criteria to place concrete barriers instead ofsteel ones. The most frequent answers were:- maintenance of barriers;- narrow median;- high volume traffic;- environment aspects.
Results from European countriesBoth steel and concrete barriers have advantages and disadvantages. The French study givesthe most detailed accident information concerning the difference between steel and concretebarriers. Mentioned is that the severity of accidents against concrete barriers is worse incomparison with accidents against steel barriers. But the amount of rollover accidents as a
77
result of a collision with a concrete barrier, is relatively high. It is know (from the UnitedStates) that accidents with a concrete New Jersey barrier with small vehicles give a highpercentage of rollovers. Rollovers are related with serious injuries, particularly if the seat beltswere not used.There are possibilities to prevent this kind of accidents by the choice of a steeper profile insteadof a special profile like that of a New Jersey barrier. It could be possible that if in the French study the number of rollover accidents is estimated asthe same for steel and concrete barriers, the difference in accident severity is approximately thesame for steel and concrete barriers. An additional analysis with the accident data is recom-mended.
In Germany the choice for steel or concrete barriers for the medians of motorways with a hightraffic flow, is based on the prevention of the collapse of the barrier. The risk for an accidentwith oncoming traffic is too high. Compared to the measures necessary to repair the damagedsteel constructions on locations where the accident risk is high, the choice for a concrete barriergives advantages. As criterium for a high traffic flow, 50,000 vehicles in 24 h. was mentioned.Concrete barriers are recommended if the width of the median is too small for the installation ofa steel barrier (< 3 m).In other situations a case by case decision should be made for the best solution. Aspects to betaken into account are: installation costs, life span, repair costs, load for bridges, risk for anaccident, difference in level between both roads, combination with a sound barrier, etc. In the German article, a table is given with a score for steel and concrete barriers for all theseaspects. It concerns here only the medians on motorways with a high traffic flow. Accordingthe total results, the difference between steel and concrete is not large. But for specific loca-tions, sometimes steel is considered for installation and in other situations concrete.In a German standard, a raised height for barriers is prescribed in the case that there is a dangerfor overriding: a height of 1.15 m instead of the normal height of 0.81 m.
In Austria it is stipulated that concrete barriers are at least equivalent to steel barriers.Mentioned is that concrete barriers give almost a full protection to the collapse of the barrier incases of accidents with cars and trucks. Experiences have shown that since the application ofconcrete barriers both the number and severity of the accidents, as well as the extent of materialdamage has decreased. The reason for the decrease of the severity is due to the reduction ofaccidents related to the collapse of barriers. Regarding the rebound at a collision, experienceswith concrete barriers are more favourable than expected.
Also in the Netherlands, steel barriers were compared with concrete barriers. Aspects takeninto account were: installation costs, costs of maintenance, and number and severity ofaccidents. The data of accidents were gathered from other countries owing to the little use ofconcrete barriers in favour of steel ones.The conclusion is that steel barriers in principle can be preferred. The total costs considered arehigher for concrete barriers in comparison with steel barriers. Concrete barriers arerecommended in situations with medians with a small space.
78
4.1.4 Proposals for standards and strategies for EU countries
Motorways: shouldersThere are safety reasons for favouring wide obstacle-free zones. 6 of the 13 countries thatprovided information maintain a minimum width of 9 m. This width is recommended as aprovisional minimum. This distance is not chosen in advance as an absolutely safe width. But bygradually widening the width of the obstacle-free zone, the road authority will be more easilytempted to choose a safety barrier than a safer obstacle-free zone. Recommended is to carryout accident investigations in different European countries to collect more data in order to takea more well-founded decision for the European situation.Slopes may be a part of an obstacle-free zone if vehicular manoeuvres are possible. This is thecase with a gradient of at least 5:1 for high slopes (> 5m) and 6:1 for lower slopes (< 2m). Onlyfixed roadside objects can be located within an obstacle free zone, if their support poles areflexible. If solitary rigid obstacles cannot be relocated, protecting them with a crash cushion isthe solution.A decision model is described by Schoon (1998) for determining the choice: obstacle free orsafety barrier.
Motorways: mediansIn terms of safety, an obstacle free-zone for off-the road vehicles in the median had to be atleast 20 m. Normally this width is out of the question. As a result, a choice has to be made forthe type of safety barriers. In this case the question is: what should be the containment level ofthe safety barrier? The level depends on the circumstances of the cross-section, the trafficvolume, proportion of heavy traffic, and so on. Some European countries have made abeginning with criteria in their standards about containment level. Drafts are being prepared. The question ‘steel or concrete barrier’ had to be answered in relation to the choice for a highor low containment level. Besides the aspect of sufficient resistance, the aspect of enoughheight to prevent rollover to the other carriageway is also important.If a decision is made for a low containment level, steel barriers are in favour if only theinstallation costs are calculated. Taking into account other aspects, it depends on the localcircumstances which type of barrier is to be preferred. Differences between countries are toogreat for a general statement here.
Express roads: dual carriagewaysIf express roads have dual carriageways, it is necessary for road safety that a safety barrier bebuilt in the median. As far as the requirements for the safety barriers are concerned, there is notmuch difference between those for safety barriers on motorways. The containment level isprobably lower because of the lower design speeds of the road.Obstacle-free zones are preferred for the shoulders. For the width of these zones, Schoon(1998) indicates that they should be about 6 m. Six of the thirteen countries who answered theSAFESTAR questionnaire use this (or a greater) width. If there is not enough space in theshoulders, safety barriers can be placed.
Express roads: single carriagewaysIt is to be preferred that the roads of this type with a high traffic volume have some kind ofphysical separation of carriageways on the road axis. This is in agreement with the (modern)ideas of separating two traffic flows in opposite directions (sustainable safety).
79
The shoulders are obstacle-free with a width of at least 6 m (see above). Safety barriers such asthose used on motorways do not fit this type of road because there is the danger of reflections,and therefore the chance of a frontal accident with traffic on the other lane. Specialconstructions are needed for this which will keep hold of a car during a collision. At thismoment a construction is being developed in the Netherlands. If there are high-risk zones (viaducts, watercourses, ravines), standard safety barriers must beplaced.
4.1.5 Strategies for attention to obstacle accidents
Renewed attention in the USIt is striking that experts in America observe that, for decennia already, attention has been paidto the problem of obstacle accidents. This problem, however, is getting bigger and bigger.Furthermore, they maintain that studies carried out in the 1970s have lost their validity manyyears ago, partly as a result of changing cars.The problems being signalled are mainly those of highways. The strategy developed in Americato deal with these problems appear to be applicable in Europe:- better accident monitoring;- research on the interaction of aspects from roads, vehicles, and drivers;- give the problem fresh attention by education, spreading information, good management,
planning;- greater budgets; working at fund-raising;- traffic laws.
Increase of obstacle problems in EuropeThe single carriageway roads are in fact at the heart of the problem of obstacle accidents inEurope. There are so many such accidents because there are so many old roads. Unfortunately,such accidents are widely spread so that dealing with them cannot be targeted at concentrationsof dangerous locations.The relative number of such accidents is also on the increase: the percentage of fatal off-the-road accidents, in comparison with the total number of accidents was 13% in 1971, and in 1996this was 26 %; a doubling during a period of 25 years. This increase applies to the Netherlands,but probably also applies to other European countries.
Identification and approachSchoon (1998) has described a procedure for identifying the locations and establishing prioritiesfor those most requiring the placing of safety barriers. Seven steps are distinguished; one ofwhich is carrying out observations of locations and a cost-benefit analysis. Owing to theapplication of a cost-benefit analysis, within this method also the ‘one million ECU test’ of theEuropean Commission can be applied.Where (isolated) fixed objects are standing in the shoulder, there are solutions for making polesetc. frangible.The ‘natural’ obstacles such as trees present a greater problem because cutting them down isprohibited and/or because of landscape preservation. Apart from the erection of safety barriers,to increase the safety of such roads, the speeds driven will have to be drastically reduced.Subsequently this means that the road's function will be changed; from that of a through-roadto one with a more local character.
80
Computer simulationsComputer simulations techniques can be used to augment the traditional crash testingprogramme, with as advantages, reducing the costs and improving the range of test conditionwhen developing or redesigning roadside hardware. At this moment the CEN standards offerno possibilities to use mathematical simulations as an instrument to support full scale results.Recommended is to discuss this item within the CEN consultations.
81
5. TUNNELS
5.1 Introduction
Road tunnels have been applied to cross natural obstacles, like mountains, rivers, canals and major navigable waterways. More recently, tunnels are applied in dense populated areas. Inthese areas the availability of land for surface routes is decreasing, while on the other handthere is a growing demand for mobility. Another development favouring the tunnel option hasbeen the increased demand for environmental protection from traffic. This includes theaesthetical value of the landscape, noises and pollution produced by large traffic streams.
The tendency is to build more complicated infrastructure with high capacity under ground.Several European countries are working on long stretches of motorways in tunnels in the nearfuture. This can be illustrated by the following two examples.
1 The Swedish ‘Ringenprojektet’. an underground motorway network to replace themotorways around Stockholm.
2 An initiative in the Netherlands, where because of new residential areas the motorwaywest of Utrecht will be built underground.
Due to technical and financial constraints, the design of tunnels on motorways is often differentfrom the design of standard motorways. An example of a financial constraint is the applicationof an emergency stopping lane. The aim of an emergency stopping lane is as follows (PIARC,Technical Tunnel Committee, 1987):I Improving safetyII Allowing vehicles to stop and allowing for breakdownsIII Allowing emergency services access to accidents on busy routesIV Allowing accidents to be bypassedV Allowing generous facilities during maintenanceVI Maintaining good level of service.
Furthermore the Technical committee Road Tunnels of PIARC wrote in their contribution tothe XVIII th World Road Congress (PIARC, 1987, p.126): "In tunnels, where the cost isextremely high for every metre extra width, care has to be taken that only economical crosssections are used in designs. Thus the question of width of tunnel emergency stopping strips orlanes has to be considered in detail.".
The problem is how one determines a solution to be economical. For a particular tunnel theextra cost for building an emergency lane can be calculated. On the other hand it is difficult tocalculate the revenues, particularly when safety is concerned. In the first place it is hard toestimate the number of saved victims, secondly valuation of the saved victims is even morecomplicated.
The objective of work package two of SAFESTAR has been special dedicated to the supportof design standards for tunnels on motorways. To guarantee safety in tunnels, it is necessary toassess to what extent it is acceptable or even required to deviate from standard motorway
82
design criteria. Furthermore, additional criteria are needed for matters affecting safety, whichonly apply to road tunnels like wall pattern and tunnel height. The study has been divided intothree parts: a literature review (Martens & Kaptein, 1998a) , a survey of tunnel designguidelines (Martens & Kaptein, 1998b) and a validation of three design features in a simulator(Martens, Törnros and Kaptein, 1998). The two partners participating in this work packagewere VTI from Sweden and TNO Human Factors Research Institute from the Netherlands.
The setup of this chapter is as follows. In the next paragraph the unsafety of motorway tunnelsis compared with the open situation. Only very rough data, retrieved from literature, is used.Additional data collection was beyond the scope of this project. The subsequent paragraphdeals with the relationship between traffic safety and alignment and cross section. Clearly,traffic safety is not the only factor which determines the alignment and the cross section of atunnel. Alignment and cross section are also the result of technical considerations like:- the ground conditions (rock, soft ground);- construction method (drill and blast, cut and cover, tunnel boring machine);- available area;- the existing road network.
In paragraph 5.4 entries and exits within a tunnel are discussed. In paragraph 5.5 some remarksare made concerning lighting in tunnels. The last paragraph of this chapter contains theconclusions and recommendations concerning future research.
Beside the subjects mentioned in the previous sections, there are several other objects in tunneldesign concerning safety which are not treated here. Tunnels, particular on motorways, areequipped with al sorts of additional equipment regarding safety. Systems like emergencytelephones, emergency exits, fire detection, water hoses, fire extinguishers, Closed CircuitTelevision (CCTV), Variable Message Signs (VMS), ventilation and drainage system have incommon that there is no direct influence on the behaviour of drivers. These systems arenecessary to keep the tunnel accessible to the public and are used once an accident hashappened. An exception is CCTV. The tunnel operator can use information collected withCCTV to close a lane, or change the mandatory speed limit in the tunnel.
5.2 Tunnel safety compared to open road conditions
In the same publication cited above the PIARC Tunnel Committee suggests that a comparisonof the tunnel situation with a simular setup in the open is useful, especially as much more dataare available on the use of the open road. On way to compare tunnels is to study accident ratesAt several World Road congresses, The PIARC Tunnel Committee has published breakdownand accident statistics of tunnels. From the statistical point of view there are several problemswith these figures. For example, several countries have their own definition of a fatal accident.In the Netherlands an accident is recorded to be fatal when a victim dies within 30 days afterthe accident, in Belgium the victim should have died on arrival in the hospital. In the followingtable the number of accidents with injured people per 108 vehicle kilometres in a number oftunnels on motorways is compared with the same figure for motorways in the open for thecountry where the tunnel is situated. The tunnel data have been extracted from the PIARCpublication “Road Safety in Tunnels” (PIARC 1995). The tunnels presented in table 1, areunidirectional tunnels, with a minimal length of 1 km, and known injury accidents rate. The data
83
of the open motorways have been extracted from the IRTAD database. The presented rate foropen roads has been calculated as the mean value of the rates of the same years studied byPIARC. In the case that IRTAD did not contain information over a simular time period aspresented by PIARC, a rate has been calculated over a time period as close as possible.
injury accidents/108 veh.km
tunnel country length[km]
lanestub.x lan.
gradient%
tunnel (studiedperiod)
open roads
OsloTunnel
Norway 1,8 2 x 3 4 - 7V shaped
7 (90-93) -
Floyfell Norway 3,9 2 x 3 1 5 (88-91) -
BrooklynBattery
USA 3,2 2 x 2 - 50 (89-91) 23 (89 - 90)
QueensMidtown
USA 2,8 2 x2 - 81 (89-91) 23 (89 - 90)
Holland USA 2,6 2 x 2 V shaped 33 (87-91) a) 23 (89 - 90)
Lincoln USA 2,5 3 x 2 V shaped 26 (87-91) 23 (89 - 90)
Elbe Germany 2,7 3 x 2 V shaped 30 (87-91) 18 (90)
Dullin France 1,5 2 x 2 2,4 4 (84-91) c) 10 (85-91)
a) multiple accidentb) with 9 entrance ramps and 9 exits rampsc) only one event was recorded
Table 5.1 Characteristics of different tunnels and their safety record
From the data in table 5.1 one cannot conclude that the safety is better in the open than intunnels nor that safety is better in tunnels than on open roads. The PIARC Tunnel committeestate however (PIARC, 1995) that road safety in tunnels is better than on open roads, except incase of failure in the geometric design. In literature (PIARC, 1983) high accident rates inexisting tunnels are associated with steep gradients and sharp curves.
The level of unsafety differs for each tunnel. The used sources provide not sufficiently data toanalyse the cause of these differences. For example, detailed data like the presence of anemergency lane.
84
5.3 Alignment and cross section
The design guidelines of the alignment and the cross section of motorways, are partly based onresearch, partly on experience. European countries have developed design guidelines more orless independently from each other, hence the design guidelines for motorways differ betweenEEC member countries. So, when we discuss the substandard design of motorway tunnels inEurope, there is also a wide variance between the tunnel design guidelines used in Europe. Indeliverable 2.2 of SAFESTAR (Martens & Kaptein, 1998b) the motorway guidelines ofSweden and the Netherlands are used as reference for the research work.
An important criterion for the design of alignment and cross section for motorways in both theopen as in tunnels is sight distance. Sight distance is an objective criterion which can becalculated from the dimensions of the road section. Because of ceiling and walls, sight distancesin tunnels are limited in comparison to the open road situation. Clearly, parameters like width,height, sight distance, applied radius in curves and gradients contribute to direct the tunnel userto behave as desired. However even more important is how the tunnel user perceives the tunnelenvironment. The perception of the tunnel user can be influences by applying certain wallpatterns, particular road markings and signing.
A limited sight distance in motorway tunnels can partly be compensated by supplying the driverwith information concerning the current traffic situation in the tunnel. This information can begiven by means of VMS, green arrow/ red cross panels. The tunnel operator can useinformation obtained from CCTV and traffic flow measurements to decide on the contents ofthe messages to the drivers.
Special attention should be paid to road sections in the vicinity of the tunnel portals. In thesesections the transition takes place from open road conditions to tunnel conditions and viceversa. On road sections close to the tunnel portals, relatively more accidents happen then in thetunnel itself (Admundsen, 1994). One major problem in the transition area is lighting (Seeparagraph 5.4). Furthermore, particular in the case of long tunnels, the cross section willchange. The road sections just outside the tunnel portals should be used to prepare the driverfor the changes he will face. Research indicates that (Martens & Kaptein, 1998a), drivers payextra attention to the tunnel entrance. It should therefore be avoided to erect signs over alength of 150-200 m in front of the tunnel
5.3.1 Alignment
In this section tunnel length, curves and gradients, will be discussed. A number of authorsreferenced by Martens & Kaptein (1998a) found by means of questionnaires that peopleexperience more fear with increased tunnel length. There are however no clues presented fromaccident statistics that a longer exposure to fear leads to more accidents.
Small curves should be avoided, especially if they are connected to a straight alignment. ThePIARC Tunnel Committee recommend observing a minimal curvature of 550-600 m. (PIARC,1995). It is not clear if this figure applies for all tunnel types. Still, a radius of 550m issubstantial less than the minimum requirement of 750 m in the Dutch motorway guidelines. A
85
radius of 750 m may only be applied in case where room is limited. Under normal conditions aradius of 2000 m. should be observed.
In case of a tunnel however there can be several other boundary conditions which forcedesigners to apply a smaller radius then one should do from road safety point of view. In thecase of a curve in a motorway in the open, the designer has several tools to inform the driverabout the curve he approaches. In tunnels it is more difficult for the driver to detect curves, topredict the radius and therefore to choose the appropriate speed. A method to inform the driverfor an upcoming curve is to use a certain wall pattern. In work package 2.3 (Martens, Törnrosand Kaptein, 1998), a simulator study with the VTI driving simulator has been performed, toinvestigate if certain wall patterns on the last 200 m of the straight section before the curveinfluence the speed behaviour of drivers. Seven different patterns, including no pattern, and asimular road stretch in the open have been studied. Subject had to drive all stretches fourtimes, of which half without speedometer readings. Driver behaviour was studied by analysingaverage speed, average lateral position, and lateral position variability. Additionally, position ofgas pedal release, position of deceleration initiation, position of maximal deceleration, andmaximal deceleration have been analysed. No influence of the simulated wall patterns wasfound however on speed behaviour. However, the results of the questionnaire show that amajority of the subjects indicate that the patterns had speed reducing effect. For three of thepatterned conditions, a small difference occurred of lateral position compared to the open roadcondition.
Another important aspect of alignment is the gradient of the road. As in the open situation, alsoin tunnels gradients influence both the capacity and road safety. Main cause is the speeddifference between cars and heavy trucks leading to inhomogeneous traffic. Because of the highcosts, steeper gradients are accepted in tunnels then for motorways in the open. More researchis needed to determine what gradient is acceptable concerning capacity and road safety fortunnels on motorways.
5.3.2 Cross section
The overall shape of the cross section is determined by the construction method. A circularcross section is the result of the use of a tunnel boring machines. The traditional horseshoe-shaped tunnel is typical for drill and blast, a square cross section one will find in cut and covertunnels.
As mentioned earlier in this chapter, a tunnel designer will whenever possible minimise the areaof the cross section, because of the high costs. The effect of the smaller cross section is thatthere is no room for an emergency lane. A number of studies have been performed to examinethe effect on behaviour when a driver drives from a motorway section with an emergency laneto a tunnel section without an emergency lane. The results of these studies show that a decreaseof the lateral width affects speed behaviour and the lateral vehicle position.
In the TNO simulator study (Martens, Törnros and Kaptein, 1998) performed in the frameworkof this study, driver behaviour was studied of subjects driving over the transition from a a widecross section into a smaller cross section. The wide cross section was according the guidelineof Dutch motorways. Four small cross sections were included into the study design, with
86
different emergency lane widths, varying from 5,75 m (emergency lane 3.50 m + 2.25 m lateralclearance) to only a lateral clearance of 0.5 m. . These four transitions were simulated in a openroad condition (control) and a tunnel condition (experimental) each for two traffic flowconditions. Driving behaviour has been expressed in terms of speed and the distance betweenthe right road marker and the right side of the car. The experiment shows that lateral width perse influences driving behaviour and that this effect is stronger in a tunnel. However, the effectson lateral position and speed are relatively small. The authors (Martens, Törnros and Kaptein,1998) recommend not to omit an entire emergency lane in tunnels. If for some reason it isimpossible to include an entire emergency lane in the design, an emergency lane of 1.50 isexpected to reduce the negative effects on road capacity and traffic safety.
5.4 Entries and Exits
In the near future the number of tunnels containing entries and exits will increase. In workpackage 2.3 (Martens, Törnros and Kaptein, 1998), the results were presented of aninvestigation into the effect of sight distance, presence of an emergency lane and traffic flow ondriving behaviour on motorway entries and exits in tunnels. The study was performed with theTNO driving simulator. Driving behaviour was measured by means of the driving speed,accepted gap, Time-To-Accident and the amount of space subjects use to perform themanouevre. For the investigated sight distances (84m-300m) no effect was found on drivingbehaviour. Furthermore, no differences were found between merging/exiting in open roadconditions and in tunnel conditions. All subjects managed to perform the manoeuvres in arelatively safe manner, also in the absence of an emergency lane. However, the results of thequestionary indicate that an emergency lane is required in case of entries in a tunnel.
5.5 Lighting
Tunnel lighting is an important issue, and there has been published quite a lot on this subject.This however has not resulted in one internationally excepted method to calculate lightinglevels in the transition area and within tunnels itself. Maybe a comparison should be madebetween the methods by evaluating the road safety in tunnels where different calculationmethods have been applied.
5.6 Conclusions and recommendations
In the ideal situation, a tunnel in a motorway should not have effect on the level of service androad safety. In reality however this is not the case. For the motorway in the open detaileddesign guidelines are developed, where tunnel designs depend largely on the local situation
Hence there is a need to quantify the effect of tunnel design parameters on traffic safety.Ideally, the tunnel designer should have a quantified relationship between a particular designfeature and road safety at his disposal to judge if a certain tunnel design is acceptable in termsof road safety.
Simulator studies give only a comparison between how drivers behave in tunnels andmotorways in the open under certain conditions. Driver behaviour in this case is described byspeed, lateral position of the vehicle and steering frequencies. There is not a direct relationship
87
between these parameters and traffic safety, i.e. registered number of accidents. Clearly,simulation is a very powerful tool to study new tunnel designs, but may be more attentionshould be paid to the relationship with the real world. A research setup would be to simulate anexisting tunnel with known accident statistics. By this way a relationship can be found betweenmeasured human behaviour and (certain types of) accidents.
Beside simulation, more structural and detailed data collection, concerning driver behaviourand accidents should be performed in existing motorway tunnels. The problem is that one needlong collection periods, al least five years (PIARC, 1995) to obtain significant results from thecollected data. During such periods there should be not to many changes in the tunnel.
In the simulator study (Martens, Törnros and Kaptein, 1998) the effect of a number of wallpatterns was examined on the speed behaviour of drivers. The idea was to reduce speed beforethe beginning of a curve. However no effect was found on speed behaviour, the basic conceptis very interesting. If for some reason the tunnel design deviates from the motorway standard,one could look for compensation by means of using the additional equipment, which is availablein tunnels but not on motorways. For example, the tunnel operator using CCTV can close alane when a vehicle has stranded. This can compensate for the absence of an emergency lane.
88
6. JUNCTIONS AND INTERCHANGES
6.1 Express roads
Though junctions and interchanges constitute a very small part of the express road network, asubstantial part of the accidents happen here. Analysis of accidents on express roads in Portugal(Cardoso & Costa, 1998) showed that on dual carriageway express roads 11 per cent of theaccidents were located at junctions; on single carriageway express roads 20 per cent of theaccidents were located at junctions. The majority of junction accidents both at dual carriagewayand single carriageway express roads are lateral accidents, and as such the express roads arecomparable to ordinary roads (see Table 6.1), where the most important junction accident typeis also a lateral accident. At dual carriageway express roads lateral accident are relatively lessfrequent than at single carriageway express roads.
Accident HITPEDESTRIANS
RUN OFFTHE ROAD
TOTALFrontal Rear end Lateral Obstacle
2x2 Motorway 5 23 16 17 4 31 96
2x2 Express road 19 12 32 11 5 18 97
2x2 Ordinary road 7 8 56 3 16 7 97
2x1 Express road 16 12 49 7 6 7 97
2x1 Ordinary road 24 11 46 5 8 5 99
Table 6.1 Accidents at junctions at different road classes by accident type (in percentages).
Given the high accident risk at junctions, in combination with the function of express roads(long distance travel) the number of interchanges should be limited. Both single and dualcarriageway express roads can have grade-separated or at-grade junctions, although grade-separated is more common at dual carriageway express roads. In general, grade-separatedjunctions are safer than at-grade junctions (Ogden, 1996). For example, Hedman (1990, citedby Ogden, 1996) found that in Sweden, grade separations resulted in 50 per cent accidentreduction at a cross junction and 10 per cent reduction at a T-junction. Whereas it is oftenheard that grade separated junctions at non-motorway roads may confuse road users, resultingin motorway driving behaviour, no empirical proof has been found that this actually happens oradversely affects safety. It is believed that the safety gains of grade-separated junctionsoutweigh the safety losses caused by possible confusion. It is very important, though, thateither type is used consistently along a particular road in order not to violate drivers’expectations and hence induce inappropriate behaviour. In a UK study on class A roads, which are very similar to express roads although they are opento non-motorised traffic, it was found that the following aspects of grade separated junctionswere associated with accident frequency (Hughes, Amis and Walford, 1996):
Grade separated junctions, on-ramps
89
1. Minor road traffic flow: a 1,000 vehicle increase in the number of vehicles entering themain road from the minor road results in a 16 % increase in accidents.
2. Vertical alignment of on-ramp: compared to an on-ramp arrangement where the verticalalignment is level, on-ramps with positive vertical alignment and negative verticalalignment are associated with increases in accident frequency of 350 % and 250 %respectively. On-ramps with a sag or crest profile are associated with 500 % moreaccidents than those with level alignment.
3. Distance to next junction: as the distance to the next junction increases, accidentfrequency at the preceding junction decreases. A one kilometre increase in inter-junctiondistance results in a 26 % decrease in accident frequency.
4. Verge width on offside of on-ramp: the presence of a wide verge on the offside of the on-ramp results in a 90 % reduction in accident frequency (compared to a narrow verge).
5. On-ramp merging length: accident frequency decreases as the length of the merging lanebetween the end nosing and the end of the on-ramp increases. A 100 m. increase inmerging length results in a 6 % decrease in accident frequency.
Grade separated junctions, off-ramps1. Exit traffic low: a 1,000 vehicle increase in the number of vehicles leaving the main road
onto the minor road results in a 13 % increase in accidents.2. Vertical alignment of off-ramp: compared to an off-ramp arrangement where the vertical
alignment is level or negative, off-ramps with positive or crest vertical alignments areassociated with an increase of 124 % in accident frequency.
3. Distance to next junction: as the distance to the next junction increases, accidentfrequency at the preceding junction decreases. A one kilometre increase in inter-junctiondistance results in a 61 % decrease in accident frequency at the off-ramp.
4. Verge width on offside of off-ramp: the presence of a wide verge on the offside of theoff-ramp results in a 79 % reduction in accident frequency (compared to a narrow verge).
Since the number of grade-separated junctions incorporated in this study was relatively small,the actual numbers should be used with care. However, it becomes clear that, in order tooptimise safety of grade-separated junctions, on-ramps and off-ramps should be level andshould have wide verges. Merging lanes need to be sufficiently long and junctions should notbe located too close to each other.
With respect to at-grade T-junctions at single and dual carriageway ‘A’ class roads (Hughesand Amis, 1996; Hughes, Amis and Walford, 1996) the following aspects were found to berelated to accident frequency:
T-junctions, single carriageway roads1. Major road traffic flow: within the major traffic flow range considered in the model
(4,500 to 17,400 vehicles per 16 hour day), an increase of 1,000 vehicles per day resultsin a 6 % increase in accidents.
2. Minor traffic flow: an increase in minor traffic flow from one categorical level to the nextresults in an increase in accident frequency of 87 %. (levels: 0-1,000 veh., 1,000-2,500veh., 2,500-4,000 veh., and 4,000-5,000 veh. per 16 hour day).
90
3. Carriageway width: within the carriageway width range considered in the model (7.0 m.to 21.2 m.), a one metre increase in carriageway width at the junction results in anestimated accident reduction of 5 %.
T-junctions, dual carriageway roads:1. Minor road traffic flow: an increase of 1,000 vehicles per 16 hour average annual
weekday flow entering the dual carriageways from the minor road results in an increaseof 120 % in accidents at the junction.
2. Gap in central reservation: T-junctions served by a gap in the central reservation areassociated with 270 % more accidents than T-junctions having no gap.The larger amount of accidents in this situation is explained by the presence of leftturning vehicles from the side road to the express road (right turning in the Britishsituation). This relatively high-risk traffic stream is excluded in the situation where thereis no gap in the central reservation.
3. Traffic using gap in central reservation: a 10 % increase in the proportion of minor roadtraffic flow using a gap in the central reservation results in a 9 % increase in accidents atthe junction.
With respect to single carriageway at-grade junctions it is difficult to come to specific designrecommendations. For dual carriage-way T-junctions it is clear that a central reserve allowingfor left turns (right turns in the UK) substantially increases the accident frequency.
Roundabouts are a very safe way of junction design since they reduce both the accident speedand the impact angle (Slop et al., 1996). However, given the high speed requirement of expressroads, they are less suitable for this type of roads. Nevertheless, as is the case in France,roundabouts could be used at the beginning and the end of an express road coming from orentering an ordinary road. This may be useful since it clearly marks the transition from one typeof road to another. In case of leaving the express road it may also contribute to breaking thehigh speed habituation process and the resulting underestimation of the own speed, which hasbeen shown to appear when people have driven at high speeds during a longer time period(ETSC, 1995).
6.2 Design of major urban junctions
SAFESTAR deals with the roads in the TERN, and this road network is situated outside urbanareas. From the TERN report it can be understood that urban sections will virtually not be partof this network in the long term. Nevertheless, in the additional information received, urbanconditions are referred to several times. In view of this, it has been assumed that urban sectionswill occur on TERN links at least for some more considerable time, presumably as mainthoroughfares. Therefore, a limited amount of attention is paid to urban conditions. Thisattention has been concentrated on major urban junctions.
6.2.1 Design principles
The design philosophy in Chapter 1 defines three design principles, which should also beapplied to the design of (major) urban junctions:
91
- A functionally planned road network: each link fits well into the whole system and actualroute choice is in accordance with planned route choice.
- A homogeneous use of the road: road users should only be confronted with smalldifferences in speed and mass.
- A recognizable road environment which stimulates the right expectations: predictabilityof traffic situations.
Elaborating these principles when classifying the road network has an important drawback forthe selection of the type of junctions:- Each road class should have a limited number of different types of junctions- A road class should only be connected to another road class according to table 6.2:
at-grade, no specificpriority regulation, speedreduction
Source: CROW (1997)
Table 6.2 Operational requirements for connecting road classes (by a certain type ofjunction)
Type of junctionSpeed reduction near and at the junction plays an important role in meeting the second designprinciple. Speed reduction can be attained by physical measures (including the application ofroundabouts), and partly by signalization (e.g. by signal coordination at successive junctions).See Table 6.3 for the type of junction which is recommended for each possible connection ofthe different road classes. The connection of two interconnective roads will mostly not besituated in the urban area, while the connection of two access roads will not be part of theTERN network
Generally four-arm junctions are not recommended for priority junctions: the number of injuryaccidents at four arm junctions is relatively high; see Figure 6.1. Roundabouts are superior toboth three- and four-arm junctions with respect to the number of injury accidents (VTI, 1998;Van Minnen, 1990; Stuwe, 1991). VTI (1998) has found the ‘best choice’ regarding type of junction for different combinations ofentering flows on the major road and the minor road (see Section 6.2.2).
An important, but not surprising, finding in many accident evaluations is that the flow level isthe most dominant predictor of the number of (injury) accidents at junctions.
roundabout, three-armsignalized or priorityjunction
Acces road n.a. roundabout, three-armsignalized or priorityjunction
three-arm junction
Source: CROW (1997)
Table 6.3 Types of junctions according to requirements in Table 6.2
Source: CROW (1995)
Figure 6.1 Relative number of accidents at four types of urban junctions
Predictability of traffic situationsThe predictability is a combination of expectation (what could happen), and observation (whatcan be seen).The observation can be improved by making the road environment less complex, e.g. byremoving obstacles which prevent a good sight, and by separating different types of conflicts(in time or space). The expectation is for a great part a matter of education and training. The road environmentcan support and ease this training by offering layouts which are as uniform as possible.Furthermore the road environment should trigger the right expectations: First of all, theapproach of a junction must be stressed. Second, the possible types of encounters must be clearfrom the marking, signing and other clues before entering the junction. Finally, the layout of thejunction must be logical and adapted to the skills of the road user, preferably the less vital roaduser.
Safety effects of design features The urban area shows an enormous variety in design elements, in layout of these elements, andin traffic situations. Evaluating the safety effects of different types of elements or layout
93
configurations, both in different traffic situations, is a huge operation. This can only be done byusing the best research methodologies, by gathering large amounts of data, and by applyingsophisticated statistical techniques. A good example of such an approach is given by Elvik et al.(1997). It will take many years to evaluate the large number of possible design elements andlayout configurations, assuming that this job will start one day.For the time being, knowledge about the safety effects of design features is spread over manyreports and institutes. Part of this knowledge is filtered through in SAFESTAR (Danish RoadDirectorate, 1998).
Priority junctionsThe relationship between the gap acceptance by motorists and the number of accidents has notbeen established yet. However, this relationship can indirectly be found by using accidentmodels which account for both the flow on the major road and the minor road. These modelswill be treated separately in this chapter (Section 6.2.2).
Source: BASt (1992)
Table 6.4 Different types of facilities for cyclists at different types of junctions
94
Facilities for cyclistsFacilities for cyclists have a great variety in design and layout. Some of these facilities havebeen evaluated thoroughly. BASt (1992) gives an overview of the most common facilities atjunctions; see Table 6.4.Facility types 20 and 21 (‘no facility’ and ‘cycle lane’) had the lowest accidents rates perjunction (also when taking the number of passing cyclists into account) compared to facilitytypes 22, 23 and 24 (‘separated cycle path’) (BASt, 1992).The Danish Road Directorate (1994) has evaluated some new types of facilities for cyclists.The main goal of these new types is to let cyclists and motorists be aware of a possibleencounter or conflict. So the facilities stress the position on the carriageway where cyclists aresuppposed to ride. The effect of these facilities on road safety appeared to be satisfactory.
RoundaboutsThe number of (injury) accidents as a function of the entering flow, can be calculated with anaccident model. This item is treated separately in this chapter (Section 6.2.2).
Three different types of facilities for cyclists (defined by BASt, 1992) have been compared witheach other (Table 6.4). For two indicators, the number of injury accidents per junction and thenumber of injury accidents per million bicycle kilometres, it appeared that type 50 (‘no cyclefacilities’ at the roundabout) had the lowest accident rates, and that type 52 (‘cycle path’) hadthe highest accident rates (BASt, 1992). This result has also been found by Van Minnen (1998).However, Van Minnen (1998) also evaluated a facility which had even lower accident ratesthan type 50: a roundabout with a cyclepath (like type 52) but with the restriction for cycliststhat they have no right-of-way at the crossings.
Signalized junctionsAs for the other types of junctions, the number of accidents related to traffic flows (motorvehicles, cyclists, and pedestrians) is treated in Section 6.2.2.Typical safety problems at these type of junctions are red-light running and rear-end accidents.Red-light running, especially by cyclists, may be caused by too long delays. Rear-end accidentshave to do with the predictability of the traffic situations: If the driver is aware of the junctionand the presence of signalization, then the driver will be prepared for vehicles stopping in frontof him/her.Regarding facilities for cyclists (Table 6.4), BASt (1992) has found that facilities 30 (‘nofacilitity’) and 31(‘cycle lane’) have lower accident rates than facilities 32, 33 and 34 (‘cyclepath’).
6.2.2 Quantitative relationships between accidents and traffic flows
Several countries are developing models for accidents on junctions. The United Kingdom has along tradition in this respect. Tanner reported already in 1953 about models for accidents onrural T junctions (Jadaan & Nicholson, 1992).Nowadays countries like Sweden, Finland and Denmark have much experience in fittingaccident models for junctions. Literature shows a slight preference for models regarding urbanjunctions. At least SAFESTAR concentrated on major urban junctions, as part of roads in theTERN network.
95
A � c�Q a1 �Q b
2 (20)
AR � a � (IM � Im)b� [
Im
(IM � Im)]c
(21)
Accident models which give a description of accidents on junctions mostly have the followingstructure:
A: number of accidentsc: estimated parameterQ1: number of entering vehicles per day on the major roadQ2: number of entering vehicles per day on the minor roada: estimated parameterb: estimated parameter
The Swedish model for motor vehicle acidents at urban juctions (no other type of road usersinvolved) is somewhat different from model type (1):
AR: accident rateIM: number of incoming motor vehicles from MAJOR roadIm: number of incoming motor vehicles from minor roada, b, c: estimated parameters
The most important difference has to do with the factor which introduces the proportion of thevehicles from the minor road in relation to the total incoming flow. The Swedish model hasbeen fitted for many types of junctions:
Number of arms three-arm
Speed limit (km/h) 50 70
Parameter a*10-8 b c a*10-8 b c
Priority 45 1.45 0.50 597 1.25 0.45
Signalized 317 1.20 0.10 651 1.20 0.10
Signalized, detection systems orseparate phase for vehicles turningleft
176 1.20 0.10 418 1.20 0.10
Roundabout 232 1.20 0 332 1.20 0
96
A � c�Q a1 �Q b
2 �K (22)
ARped. � 0.0201�Q 0.50motorv.�Q 0.28
ped. (23)
number of arms four-arm
speed limit (km/h) 50 70
parameter a*10-7 b c a*10-7 b c
priority 14.8 1.45 0.60 164 1.25 0.55
signalized 97.2 1.20 0.20 173 1.20 0.20
signalized, detection systems orseprate phase for vehicles turningleft
49.9 1.20 0.20 94 1.20 0.20
roundabout 30.8 1.20 0 44 1.20 0
Table 6.5 Estimated parameters in the Swedish model (type 21)
Models which are more complicated can be used to relate the number of accidents to detailedcharacteristics of the junction, e.g. different types of medians (CROW, 1997) or the number ofarms of a junction:
K: design element
Sometimes flows Q1 and Q2 are added instead of multiplied.The Nordic countries are developing models in which the influence of bicycle flows isincorporated.
Cyclists and pedestriansBrüde & Larsson (1993) have analyzed accidents at junctions which are situated in 30 Swedishurban areas with more than 25,000 inhabitants. They were especially interested in accidentsconcerning cyclists and pedestrians. Their analysis comprised 432 accidents with cyclists on377 junctions, and 165 accidents with pedestrians on 285 junctions. Only junctions withpedestrian or bicycle flows of more than 100 per day have been selected.Brüde & Larsson fitted a model of type (1), taking the entering motor vehicles as Q1 and thenumber of crossing cyclists or pedestrians as Q2. A pedestrian or cyclist who crosses two timesat a junction (e.g. if he/she has to cross two carriageways in order to turn left) is counted astwo pedestrians or cyclists. The model for the accident rate of pedestrians is:
ARped.: number of accidents with pedestrians per million crossing pedestrians ARmotorv.:number of accidents with motorvehicles per million passing motorvehicles
97
ARcyclist �0.0494�Q 0.52motorv.�Q 0.35
cyclist (24)
Acyclist �0.0000180�Q 0.52motorv.�Q 0.65
cyclist (25)
A� I ��p ��e �X (26)
The model for the accident rate concerning accidents with cyclists is:
ARcyclist: number of accidents with cyclists per million crossing cyclists
RoundaboutsBrüde & Larsson (1996) used another model for accidents with cyclists at urban roundabouts:
Acyclist: number of accidents with cyclists
This model was derived from Swedish data and was tested successfully by using Danish andDutch data.Van Minnen (1995) showed that the number of injury accidents at roundabouts is proportionalto the number of entering motor vehicles. This applies to the total number of injury accidents,and to the number of injury accidents with bicycles/mopeds.
United KingdomThe Transport Research Laboratory has recently reported about an extensive research into the(number of) accidents at different types of urban junctions:- three-arm priority junctions
Summersgill et al. (1996) used data from 980 three-arm priority junctions. The totalnumber of accidents at these junctions amounted to 2699 in a period of five years. Thebiggest part of the junctions has a speed limit of 30 mph, a smaller part has a speed limitof 40 mph.
- three-arm signalized junctionsTaylor et al. (1996) has investigated 221 tree-arm signalized junctions at which 2262accidents occurred in a six-year period.
- four-arm priority junctions and staggered junctionsLayfield et al, (1996) selected 300 junctions (2917 accidents). The junctions have either a40 mph limit or a 30 mph limit.
Many geometrical and other characteristics of each type of junction have been filed. The dataabout the vehicle flows were divided into the six possible flows at three-arm junctions and thetwelve possible flows at four-arm junctions.All fitted models are of model type (20) or of the following type:
98
A: number of accidentsI: number of entering vehicles on major and minor roadp: percentage of the flow on the minor road�: estimated parameter�: estimated parameter�: estimated parameterX: geometrical or technical element
These models require a high level of accuracy of the input data. In this case special attentionhas been paid to the inaccuracy of the flow data (Summersgill et al., 1996; page 31).Dozens of models were fitted, for the total number of accidents, for different types ofaccidents, and for types of accidents related to specific characteristics of a junction.Unfortunately the researchers have not given a clue for road designers to find their way in themany models. Much basic information is available; but a lot of effort should be undertaken tomake it applicable for the road designer.
DenmarkThe Danish Road Directorate (1995) fitted a model of type (20) using data from 1036 majorurban junctions. Table 6.6 shows the parameters a, b and c for accidents with injury, fordifferent types of junctions.
Accidents with injury Parameters in model (1)
Type of junctionsc a b
Three arms, priority 2.98·10-6 0.81 0.52
Three arms, signalized 7.04·10-8 1.36 0.32
Four arms, priority 1.68·10-4 0.36 0.58
Four arms, signalized 8.62·10-5 0.52 0.47
Tabel 6.6 Danish parameters in model type (20)
Other models have been developed taken into account several geometrical and othercharacteristics of junctions. However these more complicated models did not really show betterresults than model type (20). For this reason model type (20) will be applied.
Results from SAFESTARStarting from the Danish models of type (20) and the Swedish models (21), 23, and (24),SAFESTAR aimed at testing these models using data from other countries, viz. theNetherlands, France, and the Czech Republic. Furthermore the Danish and Swedish modelshave been compared with each other for a few combinations of input values.Finally recommendations about the application of certain types of junctions have beenformulated.
99
The Swedish and Danish modelsThe Swedish model type (21) for the accident rates of motor vehicle accidents shows that agrowing number of incoming motor vehicles results in a higher accident rate. This finding isvalid for all types of junctions (see also table 6.5); see Figures 6.2A and 6.2B.
In the models (23) and (24) respectively increasing numbers of crossing cyclists and pedestrians result in lower accident rates for cyclists and pedestrians. At the same time these models implythat the absolute number of accidents with cyclists and pedestrians is increasing when thecyclists’ and pedestrians’ flows grow (and also increasing when the number of incoming motorvehicles grows); see also figures 6.3 and 6.4.
The Danish model type (20) shows a different result compared to the Swedish model type (21).In this comparison the same input values have been used. For four-arm junctions the accidentrates of motor vehicle accidents show an accident rate which is nearly constant, independent ofthe number of incoming motor vehicles. The increasing accident rates of the three-armjunctions are comparable to the results of the Swedish model.The accident rates in these Swedish and Danish models have about the same level.
Both Denmark and Sweden are using the same type of models for accidents with cyclists(type 24). Therefore the differences in the results are small (both models show a decreasingaccident rate and an increasing number of accidents when the number of cyclists increases). The accident level for this type of accident is apparently higher in Denmark than in Sweden.
Testing the models with data from other countriesThe models to be tested have a rather simple structure, and the input seems to be of a simplenature as well. However, input data about the number of cyclists appeared to be hardly everavailable, except for some junctions in the Netherlands. So most of the testing concentrated onthe models of type (20) and (21).
Numbers of accidents with motorvehicles were available for five different types of junctions inthe Netherlands. Both these numbers and the accident rates have been compared to thepredicted values of the Danish and Swedish models (models 20 and 21). The predicted numberof accidents by the Swedish model showed a reasonably good fit, while the Danish modelconsistently resulted in much too high numbers. However, the Danish model includes accidentswith unprotected road users as well. See also Table 6.7 for a summary of the results.
Accidents with cyclists are overestimated by both the Swedish and the Danish models(type 24). Table 6.8 shows a clear overestimation for all types of junctions, except for the four-arm signalized junction with a speed limit of 70 km/h.
The French data only stemmed from four-arm signalized junctions. The description about thetype of signalization seemed to be somewhat incomplete. That is why the Swedish model hasbeen used for two different assumptions: if the signalization at the junction is of a simple type(no detection systems and no separate phase for vehicles turning left as well) then the accidentnumbers and the accident rate are predicted quite well. But if the assumption is that thesignalization is equiped with those features, then the model fits badly. Table 6.9 shows theresults.
Table 6.7 Comparison between observed and predicted number of accidents and accidentrates (motor vehicles only) for a number of Dutch junctions
The Czech junctions have a speed limit of 60 km/h. To meet this condition, the Swedish modelfor 70 km/h was used. As for the French junctions, the prediction was made for twoassumptions: signalization systems with and without detection systems or a separate phase forvehicles turning left. The best fit appeared to result from the assumption of a simplesignalization system; see also Table 6.10.
Recommendations for applying different types of junctionsBrüde & Larsson (1998) have calculated the differences between the application of differenttypes of junctions in the same conditions (number of incoming motor vehicles, proportion ofthe flow from the minor road). These differences in the number of accidents can be translatedto the costs of these accidents. If a certain type of junction will save accident costs compared toan ordinary priority junction, then the application of the safer type will be profitable whenrebuilding such a junction. The expected future accident costs of the existing type are higherthan the expected costs of the safer type. This approach resulted in the following table 6.11:
Sw. model: predicted accident rate (signals with a detection system or a separate phase forvehicles turning left)
0.087
4S50: four-arm signalized junction, speed limit 50 km/h
Table 6.9 Comparison between observed and predicted number of accidents and accidentrates (only motor vehicles involved) for a number of French junctions
102
3P60 3S60 4P60 4S60 Total
number of junctions 29 21 20 21 91
number of observed accidents 83 113 119 173 488
Sw. model: predicted number of accidents (simpletraffic signals)
78.0 79.6 121.5 210.7 489.8
Sw. model: predicted number of accidents (signalswith a detection system or a separate phase forvehicles turning left)
Table 6.10 Comparison between observed and predicted number of accidents and accidentrates (only motor vehicles involved) for a number of Czech junctions
Incoming motor vehicles per day
5 15 30
prportion of motor vehicles comingfrom minor road
5 20 50 5 20 50 5 20 50
three-arm junction50 km/h
- - - - - S+R
- S+ S+RS
four-arm junction50 km/h
- - R - RS+
RS+
S
S+R
RS+
RS+
S
three-arm junction70 km/h*
- - R - RS+
RS+
- RS+
RS+
S
four-arm junction70 km/h*
- RS+
RS+
R RS+
RS+
S
RS+
RS+
RS+
S
-: replacement is not profitableR: roundabout; S: signalized junction; S+ signalized junction with detection system or a separate phase for vehicles
turning left*: at these roads roundabouts usually have a speed limit of 50 km/h
Table 6.11 Type of junction which is recommended according to outcome of the Swedishmodel
103
Figure 6.2A Accident rates (accidents with injury) as a result of the Swedish model, forroads with a speed limit of 50 km/h
Figure 6.2B Accident rates (accidents with injury) as a result of the Swedish model, forroads with a speed limit of 70 km/h
104
Figure 6.3 Accident rates (accidents with cyclists involved) as a result of the Swedish model
Figure 6.4 Number of accidents (accidents with cyclists involved) as a result of the Swedishmodel
105
REFERENCES
1. Introduction
1.1 Design philosophy
ERSF (1996). Technical guide on road safety for interurban roads (INTERSAFE). European Road SafetyFederation, Brussels.
Fink, K.L. & R.A. Krammes (1995). Tangent length and sight distance effects on accident rates at horizontalcurves on rural two-lane highways.In: Transportation Research Record 1500. Transportation Research Board, Washington, D.C.
Köppel, G. & H. Bock (1979). Fahrgeschwindigkeit in Abhängigkeit von der Kurvigkeit. Reihe Straßenbau undStraßenverkehrstechnik. Heft 269. Bundesminister für Verkehr, Bonn.
Krammes, R.A. (1997) Interactive Highway Safety Design Model: Design consistency module. In: Public Roads.Sept./Oct. pp. 47-51.
Krammes, R.A.; K.S. Rao & H. Oh (1995). Highway geometric design consistency evaluation software. In:Transportation Research Record 1500. Transportation Research Board, Washington, D.C.
Krammes, R.A. & S.W. Glascock (1992). Geometric inconsistencies and accident experience on two-lane ruralhighways. In: Transportation Research Record 1356. Transportation Research Board, Washington, D.C.
Lamm, R.; J.C. Hayward & J.G. Cargin (1986). Comparison of different procedures for evaluating speedconsistency. In: Transportation Research Record 1100. Transportation Research Board, Washington, D.C.
Lamm, R.; E.M. Choueiri; J.C. Hayward & A. Paluri (1988a). Possible design procedure to promote designconsistency in highway geometric design on two-lane rural roads.In: Transportation Research Record 1195. Transportation Research Board, Washington, D.C.
Lamm, R.; E.M. Choueiri; J.C. Hayward (1988b). Tangent as an independent design element.In: Transportation Research Record 1195. Transportation Research Board, Washington, D.C.
Lamm R.; H. Steffen & A.K. Guenther (1994a). Procedure for detecting errors in alinement design andconsequences for safer redesign. In: Transportation Research Record 1445. Transportation Research Board,Washington, D.C.
Lamm, R. & B.L. Smith (1994b). Curvilinear alinement: An important issue for more consistent and safer roadcharacteristic. In: Transportation Research Record 1445. Transportation Research Board, Washington, D.C.
Lamm, R.; A.K. Guenther & E.M. Choueiri (1995). Safety module for highway geometric design. In:Transportation Research Record 1512. Transportation Research Board, Washington, D.C.
Leisch, J.E. & J.P. Leisch (1977). New concepts in design-speed application. In: Transportation Research Record 631. Transportation Research Board, Washington, D.C.
Lippold, C. (1996). Weiterentwicklung der Relationstrassierung von Landstraßen. In: Straßenverkehrstechnik No. 4. pp. 165-171.
Messer, C.J.; J.M. Mounce & R.Q. Brackett (1981). Highway design consistency related to driver expectancy.Volume III: Procedures for determining geometric consistancy. Publication No. FHWA-RD-81-037. FederalHighway Administration, Washington, D.C.
106
Opiela, K.S; H.W. McGee; W.E. Hughes & K. Daily (1995). Relationships between highway safety andgeometric design. Transportation Research Board, Washington, D.C.
Paniati, J.F. & J. True (1996). Interactive Highway Safety Design Model (IHSDM): Designing highways withsafety in mind. In: Transportation Research Circular 453. Transportation Research Board, Washington, D.C.
1.2 Road Safety Audits
References United Kingdom
Royal Society for the Prevention of Accidents. “Road Safety Engineering Manual”
The Institution of Highways and Transportation, 1996. “Guidelines for The Safety Audit Of Highways”, IHT.
The Institution of Highways and Transportation 1990. “Guidelines for The Safety Audit Of Highways”, IHT.
Review of Safety Audit procedures The Institution of Highways & Transportation March 1995 by MissA. Crafer.
The use of Road Safety Audits in Great Britain by S. Proctor & M. Belcher TMS Consultancy February 1993.
Safety Audit Policy Northamptonshire County Council, August 1991.
Safety Practice Note Edition 1 “Safety Audit” Kent County Council, February 1994.
References Denmark
Håndbog. Trafiksikkerhedsrevision - for hovedlandevejeVejdirektoratet, SSV, København, August 1993.Manual. Safety Audit - on Trunk RoadsThe Danish Road Directorate, SSV, Copenhagen, August 1993.
Road Safety Audit, The Danish ExperiencesSchelling, Adriaan, Road Safety and Environment, The Danish Road Directorate.Paper presented at the International Conference, Strategic Highway Research Program and Road Safety,Prague, the Czech Republic, September 1995.
Safety Audit in Denmark - a Cost Effective ActivityArticle by Wrisberg, Jacob and Nilsson, Puk Kristine, Road Safety and Environment, The Danish Road Directorate, November 1996.
Hvad er trafiksikkerhedsrevision, og hvordan fungerer det?Schelling, Adriaan, Trafiksikkerhed og Miljø, Vejdirektoratet.Dansk Vejtidsskrift, December 1994.
What is Safety Audit, and how does it work?Schelling, Adriaan,The Danish Road Magazine, December 1994.
2. Draft of a new Danish Manual of Safety Audit.Road Safety and Environment, The Danish Road Directorate. Edited by Gaardbo, Anders Møller, November 1996.
107
Trafiksikkerhedsrevisionsprojektet - Evaluering. Det eksterne panels rapport.Jørgensen, N.O. et al.Vejdirektoratet, København 1995.The Safety Audit Project - Evaluation. The External Panel’s Report.Jørgensen, N.O. et al.The Danish Road Directorate, Copenhagen 1995.
References Norway
Draft of Norwegian quality handbook - guidelines for inspection and quality audit:Trafikksikkerhet i prosjektet - kontroll og revisjon av planerRoad safety in the scheme - inspection and audit of plansAntonsen, Jan; Andersen, Øyvind; Hvoslef, Henrik; Skånlund, Anders, all from the Norwegian RoadDirectorate (Statens vegvesen Vegdirektoratet). Furthermore Rognerud, Magne, Public RoadsAdministration in Buskerud (Statens vegvesen Buskerud), and Nyland, Knut, Public RoadsAdministration in Rogaland (Statens vegvesen Rogaland).Oslo, 10th April 1996.
Kvalitetshåndbok for Statens vegvesen, Nivå A.Statens Vegvesen, mars 1996.Quality manual for the Norwegian Public Roads Administration, Level A.Public Roads Administration, March 1996.
Strategier for kvalitetsutvikling i Statens vegvesen i perioden 1996-2001, Nivå A.Statens vegvesen, Vegdirektoratet, September 1996.Strategies for quality development in the Public Roads Administration in the period1996-2001, Level A. Public Roads Administration, Road Directorate, September 1996.
Norsk Standard NS-IS 10011 1Udarbejdet af Norsk Verkstedsindustris Standardiseringssentral (NVS). 1. utg., August 1992.Norwegian Standard NS-IS 10011 1Compiled by the Standardization Central of the Norwegian Workshop Industry.1. edition, August 1992.
In addition conversations and correspondence with Knut Nyland, Statens vegvesen Rogaland (The PublicRoads Administration in Rogaland). References New Zealand
Safety audit policy and procedures, August 1993 Transport New Zealand.
Safety audit procedures for existing roads, August 1995 Draft 3, BECA Carter Hollings & Ferner LTD.
Pilot safety audits: local roading Projects. {summary} October 1994 Transit New Zealand. Report No. 94/338S
Review of the Implementation of Safety Audit on State Highways. {summary} December 1994 Transit NewZealand. Report No. 94/346S
Review of a selection of rural safety audits, July 1996 Review and Audit Division. Report No. 95/415S
Review of a selection of urban safety audits, July 1996 Review and Audit Division. Report No. 95/416S
In addition correspondence with Ian Appleton, TNZ.
108
References Australia
Proceedings 17th ARRB. conference, part 5, Geoff Middleton.
Road Safety Audit Austroads SAA HB43-1994
Road Safety Audit “its progress in Australia and New Zealand”
Philip Jordan, Ian Appleton, ITE annual meeting 1994
References USA
U.S. Department of Transportation, 1980. “ Safety Design and Operational Practices”
M. A. Wallen, 1990. “ What makes a Good Safety Management System ?”, Institute of TransportationEngineers (ITE) Journal, January 1993.
Institute of Transportation Engineers Technical Council Committee 4S-7, 1995. “Road Safety Audits: A Newtool for Accident Prevention”. Institute of Transportation Engineers (ITE) Journal, February 1995.
References France
a New approach to improving road safety: Safety checking of road infrastructure.
FERSI September 1996 Christian Machu. at the International Conference on Road Safety in Europe
Sécurité des routes et des rues September 1992, SETRA et CETUR Instructions for design and management.
1.3 The Phenomenon of express roads
Cardoso J. & Costa S. (1998) Comparative analysis of road safety on express road; report on workpackage 3.1 ofSAFESTAR. Laboratório Nacional de Engenharia Civil (LNEC), Lisbon (in preparation)
Schagen, I.N.L.G. van (1998) Express roads in Europe; report of workpackage 3.2 and 3.3 of SAFESTAR.SWOV Institute for Road Safety Research, Leidschendam (in preparation)
Hughes, W. & Amis, G. (1996) Accidents on rural roads: single carriageway ‘A’ class roads. Foundation forRoad Safety Research, Basingstoke.
Hughes, W., Amis, G. & Walford, A. (1996) Accidents on rural roads: dual carriageway ‘A’ class roads.Foundation for Road Safety Research, Basingstoke.
OECD (in preparation) Safety strategies for rural roads. Organisation for Economic Co-operation andDevelopment, Paris.
Hummel, T. (1998) Safety standards for express roads; report on workpackage 3.4 of SAFESTAR. SWOVInstitute for Road Safety Research, Leidschendam (in preparation).
109
O’Cinneide, D. (1995) The relationship between geometric design standards and safety. Paper presented at theInternational Symposium on Highway Geomeric Design Standards in Boston Massachusetts. TransportationResearch Board, Washignton, D.C.
Slop, M., Jouineau, J.P., Machu, C., Weber, R., Bergh, T., Zaragoza Ramirez, A. & Janssen, T. (1996) Technical guide on road safety for interurban roads; INTERSAFE. European Road Safety Federation, Brussels.
2.1.2 Rural roads
Cardoso, J.; A. Flouda; I. Dimitropoulos & G. Kanellaidis (1998). Design consistency of horizontal alignmentin rural roads. Laboratório Nacional de Engenharia Civil, Lisbon & National Technical University of Athens,Department of Transportation Planning and Engineering, Athens.
Cardoso, J. (1996). Estudo das relações entre as características da estrada, a velocidade e os acidentesrodoviários. Aplicação a estradas de duas vias e dois sentidos. Ph.D. Thesis. Universidade Técnica de Lisboa &Laboratório Nacional de Engenharia Civil, Lisbon
Cleveland, D.E. & R. Kitamura (1987). Macroscopic modeling of two-lane rural roadside accidents. In:Transportation Research Record 681. Transportation Research Board, Washington, D.C.
Collins, K.M. & R.A. Krammes (1996). Preliminary validation of a speed-profile model for design consistencyevaluation. In: Transportation Research Record 1523. Transportation Research Board, Washington, D.C.
ERSF (1996). Technical guide on road safety for interurban roads (INTERSAFE). European Road SafetyFederation, Brussels.
FGSV (1984). Richtlinien für die Anlage von Straßen (RAS). Teil: Linienführung (RAS-L). Abschnitt 1:Elemente der Linienführung (RAS-L-1). Forschungsgesellschaft für Straßen- und Verkehrswesen, Köln.
FGSV (1995). Richtlinien für die Anlage von Straßen (RAS). Teil: Linienführung (RAS-L).Forschungsgesellschaft für Straßen- und Verkehrswesen, Köln
FHWA (1992). Safety effectiveness of highway design features. Volume III: Cross sections. Publication No.FHWA-RD-91-046. Federal Highway Administration, Washington, D.C.
Fink, K.L. & R.A. Krammes (1995). Tangent length and sight distance effects on accident rates at horizontalcurves on rural two-lane highways. In: Transportation Research Record 1500. Transportation Research Board,Washington, D.C.
Fitzpatrick, K.; C.B. Shamburger; R.A. Krammes & D.B. Fambro (1997). Operating speed on suburban arterialcurves. In: Transportation Research Record 1579. Transportation Research Board, Washington, D.C.
Garner, G.R. & R.C. Deen (1973). Elements of median design in relation to accident occurrence. In: HighwayResearch Record 432. Highway Research Board, Washington, D.C.
Glennon, J.C. & C.A. Joyner (1969). Re-evaluation of truck climbing characteristics for use in geometricdesign. Reseacrh Report 134-2. Texas Transportation Institute. Texas A&M University.
Graham, J.L. & D.W. Harwood (1982). Effectiveness of clear recovery zones. National Cooperative HighwayResearch program. Report 247. Transportation Research Board, Washington, D.C.
Harwood, D.W.; A.D. St. John & D.L. Warren (1985). Operational and safety effectiveness of passing lanes ontwo-lane highways. In: Transportation Research Record 1026. Transportation Research Board, Washington,D.C.
110
Harwood, D.W. (1990). Effective utilization of street width on urban arterials. National Cooperative HighwayResearch program. Report 330. Transportation Research Board, Washington, D.C.
Harwood, D.W. (1986). Multilane design alternatives for improving suburban highways. National CooperativeHighway Research program. Report 282. Transportation Research Board, Washington, D.C.
Köppel, G. & H. Bock (1979). Fahrgeschwindigkeit in Abhängigkeit von der Kurvigkeit. Reihe Straßenbau undStraßenverkehrstechnik. Heft 269. Bundesminister für Verkehr, Bonn.
Krammes, R.A. (1997) Interactive Highway Safety Design Model: Design consistency module. In: Public Roads.Sept./Oct. pp. 47-51.
Krammes, R.A.; K.S. Rao & H. Oh (1995). Highway geometric design consistency evaluation software. In:Transportation Research Record 1500. Transportation Research Board, Washington, D.C.
Krammes, R.A. & S.W. Glascock (1992). Geometric inconsistencies and accident experience on two-lane ruralhighways. In: Transportation Research Record 1356. Transportation Research Board, Washington, D.C.
Kulmala, R. & Roine, M. (1988). Accident prediction models for two-lane roads in Finland. In: Stichting Wetenschappelijk Onderzoek Verkeersveiligheid SWOV, Traffic safety theory and researchmethods, April 26-28 1988, Amsterdam.
Lamm, R.; J.C. Hayward & J.G. Cargin (1986). Comparison of different procedures for evaluating speedconsistency. In: Transportation Research Record 1100. Transportation Research Board, Washington, D.C.
Lamm R. & E.M. Choueiri (1987). Recommendations for evaluating horizontal design consistencyinvestigations in the State of New York. In: Transportation Research Record 1122. Transportation ResearchBoard, Washington, D.C.
Lamm, R.; E.M. Choueiri; J.C. Hayward & A. Paluri (1988a). Possible design procedure to promote designconsistency in highway geometric design on two-lane rural roads.In: Transportation Research Record 1195. Transportation Research Board, Washington, D.C.
Lamm, R.; E.M. Choueiri; J.C. Hayward (1988b). Tangent as an independent design element.In: Transportation Research Record 1195. Transportation Research Board, Washington, D.C.
Lamm R.; H. Steffen & A.K. Guenther (1994a). Procedure for detecting errors in alinement design andconsequences for safer redesign. In: Transportation Research Record 1445. Transportation Research Board,Washington, D.C.
Lamm, R. & B.L. Smith (1994b). Curvilinear alinement: An important issue for more consistent and safer roadcharacteristic. In: Transportation Research Record 1445. Transportation Research Board, Washington, D.C.
Lamm, R.; A.K. Guenther & E.M. Choueiri (1995). Safety module for highway geometric design. In:Transportation Research Record 1512. Transportation Research Board, Washington, D.C.
Leisch, J.E. & J.P. Leisch (1977). New concepts in design-speed application. In: Transportation Research Record 631. Transportation Research Board, Washington, D.C.
Lippold, C. (1996). Weiterentwicklung der Relationstrassierung von Landstraßen. In: Straßenverkehrstechnik No. 4. pp. 165-171.
Mak, K. (1987). Effect of bridge width on highway safety. In: State-of-the-art Report 6. TransportationResearch Board, Washington, D.C.
111
Mak, K. (1995). Safety effects of roadway design decisions - roadside. In: Transportation Research Record1512. Transportation Research Board, Washington, D.C.
Messer, C.J.; J.M. Mounce & R.Q. Brackett (1981). Highway design consistency related to driver expectancy.Volume III: Procedures for determining geometric consistancy. Publication No. FHWA-RD-81-037. FederalHighway Administration, Washington, D.C.
Paniati, J.F. & J. True (1996). Interactive Highway Safety Design Model (IHSDM): Designing highways withsafety in mind. In: Transportation Research Circular 453. Transportation Research Board, Washington, D.C.
Polus, A.; M. Livneh & J. Craus (1984). Effect of traffic and geometric measures on highway average runningspeed. In: Transportation Research Record 960. Transportation Research Board, Washington, D.C.
Schoon, CC. & J.M.J. Bos (1983). Boomongevallen; Een verkennend onderzoek naar de frequentie en ernst vanbotsingen tegen obstakels in relatie tot de breedte van de obstakelvrije zone. R-83-23. StichtingWetenschappelijk Onderzoek Verkeersveiligheid SWOV, Leidschendam.
TRB (1987). Designing safer roads. Practices for Resurfacing, Restoration, and Rehabilitation. Special Report214. Transportation Research Board, Washington, D.C.
Turner, D.S. (1984). Prediction of bridge accident rates. In: Journal of Transportation Engineering. Volume110. No. 1.
Zegeer, C.V.; D.W. Reinfurt; J. Hummer; L. Herf & W. Hunter (1988a). Safety effects of cross-section designfor two-lane roads. In: Transportation Research Record 1195. Transportation Research Board, Washington,D.C.
Zegeer, C.V.; D.W. Reinfurt; W.W. Hunter; J. Hummer, R. Stewart & L. Herf (1988b). Accident effects ofsideslope and other roadside features on two-lane roads.In: Transportation Research Record 1195. Transportation Research Board, Washington, D.C.
Zegeer, C.V.; R. Stewart; D. Reinfurt; F. Council; T. Neuman; E. Hamilton; T. Miller & W. Hunter (1990).Cost effective geometric improvements for safety upgrading of horizontal curves. Volume 1. Final report.Highway Safety Research Center. University of North Carolina, Chapel Hill.
Zegeer, C.V.;R. Stewart & F. Council (1994). Roadway widths for low-traffic-volume roads. NationalCooperative Highway Research program. Report 362. Transportation Research Board, Washington, D.C.
3. Cross-section
3.2 Express roads
Armour, M. (1984). The relationship between shoulder design and accident rates on rural highways. In:Proceedings of the 12 th Conference of the Australian Road Research Board ARRB, Hobart, Tasmania, august27-31, 1984. Volume 12, part 5: Traffic Behaviour.
Bruhning, E. (1977). Untersuchung der Unfalle met Personenschaden auf Autobahnen. In: Strassenbau undStrassenverkehrstechnik, Heft 223, 1977. Bonn.
Cardoso, J. & Costa, S. (1998). Comparative Analysis of Road Safety on Express Roads. Report of Safestarworkpackage 3.1. SWOV, Leidschendam.
Garner, G.R. & Deen, R.C. (1973). Elements of median design in relation to accident occurrence. Paperprepared for TRB Annual Meeting. Transportation Research Board, Washington, D.C.
112
Hadi, M.A., Aruldhas, J., Chow, L.F. & Wattleworth, J.A. (1995). Estimating the safety effects of cross-sectiondesign for various highway types using negative binomial regression. Paper presented at the 74 th AnnualMeeting of the Transportation Research Board. Preprint paper no.: 950615, Transportation Research Board,Washington, D.C.
Harwood, D.W., Hoban, C.J. & Warren, D.C. (1988). Effective use of passing lanes on two-lane highways. In:Transportation Research Record, Serial 1988 no. 1195. Transportation Research Board, Washington, D.C.
Hedman, K.O. (1990). Road Design and Safety. In: Proceedings of Strategic Highway Research Program andTraffic Safety on Two Continents. Gothenburg, 1989. VTI Report 315A, 1990.
Hughes, W. & Amis, G. (1996). Accidents on Rural Roads: Single Carriageway ‘A’Class Roads. Foundationfor Road Safety Research, Basingstoke.
Hughes, W., Amis, G. & Walford, A. (1996). Accidents on Rural Roads: Dual Carriageway ‘A’Class Roads.Foundation for Road Safety Research, Basingstoke.
Martin and Voorhees Associates (1978). Crawler lane study. An economic evaluation. Department ofEnvironment, London.
Michalski, L. (1994). Road cross-section. Annex VI to SWOV-report Safety effects of road design standards. A-94-8. SWOV, Leidschendam.
Motorway Working Group (1994). Standardisation of typology on the Trans-European Road Network - ActionSTART. European Commission, Directorate General for Tranport, Brussels.
Oellers (1976). Untersuchung uber den Einfluss der Fahrstreifenbreite auf den Verkehrsablauf aufRichtungsfahrbahnen. In: Strassenbau und Strassenverkehrstechnik, heft 211, 1976.
Richtlijnen voor het Ontwerpen van Autosnelwegen (ROA); I Basiscriteria (1992); II Alignement (1993); IIIDwarsprofielen (1993).
Zegeer, C.V., Deen, R. & Mayes, J. (1981). Effect of lane and shoulder widths on accident reduction on rural,two-lane roads. Transportation Research Record 806. Transportation Research Board, Washington, D.C.
Zegeer, C.V. & Deacon, J.A. (1987). Effect of lane width, shoulder width and shoulder type on highway safety.In: Relationships between safety ad key highway features; State of the art report 6. Transportation ResearchBoard, Washington, D.C.
Zegeer, C.V., Hummer, J., Herf, L., Reinfurt, D. & Hunter, W. (1987). Safety Cost-Effectiveness of IncrementalChanges in Cross-Section Design. Informational Guide. Report no.: FHWA/RD-87/094. Federal HighwayAdministration, Washington, D.C.
Zegeer, C.V., Reinfurt, D.W., Hummer, J., Herf, L. & Hunter, W. (1988). Safety effects of cross-section designfor two-lane roads. In: TRR, Serial 1988 no. 1195. Transportation Research Board, Washington, D.C.
Danish Road Directorate: Forsøg med færdselssøm og profilerede striber som vejmidteafmærkning. DanishRoad Directorate, Denmark
Donald, D.: Be warned, a review of curve warning signs and curve advisory speeds. ARRB Transport Research,Australia.
Elvik, R., Mysen, A.B. & Vaa, T. (1995). Trafiksikkerhetshåndbok. TØI, Norway.
ETSC: Forgiving Roadsides.
Fildes, B.: Perceptual Countermeasures. Monash University Accident research Centre, Australia.
Kalberg, V.: Safety of roadside reflector post: Just an illusion? VTT, Finland.
Kanellaidis, G.: Factors Affecting Drivers Choice of Speed on Roadway Curves. Journal of Safety Research,Vol 26, No. 1, 1995.
Lamm, R.: Design of Motorways and rural roads with special emphasis on traffic safety. University ofKarlsruhe, Germany.
Midtland, K. & Mushaug, T.: Trafikkskilting i Norden (Road Signing in the Nordic Countries). TØI, Norway.
New York Department of Transportation: SAFE-STRIPS, NYSDOT. USA.
Plant, J.: Raised-rib road-marking, research into the safety implications. Allott & Lomax ConsultingEngineers, England.
Retting, C.M. & Farmer, C.M.: Use of Pavement Markings to Reduce Excessive Traffic Speeds on HazardousCurves. Insurance Institute for Highway Safety, USA.
RfT: Trafikanters forståelse af skiltning og Kørebaneafmærkning. RfT, Denmark.
RfT: Trafikanters forståelse af færdselstaveler (Motorist comprehension of various traffic signs). RfT,Denmark.
SETRA: Guidelines on the signing of curves, case study. SETRA, France
SETRA: Traffic Safety on Roads and Streets. SETRA, France.
SETRA: Technical Recommendations for the General Design and the Geometry of Roads - Highway DesignGuide (except for expressways). SETRA, 1994, France.
Statens Vegvesen, Norway: Enkle fysiske tiltak mot møte- og utforkjøringslykker.VIC Roads: Traffic Engineering Manual vol. 2. VIC Roads, Australia.
114
TØI: Metode for valg af visuelle virkemidler i vegtrafikken. TØI, Norway.
Zwahlen, H. & Schnell, T.: Curve Warning System and the Delineation of Curves with Curve DelineationDevices. Ohio University, USA.
4. Safety devices
AASHTO (1989). Roadside design guide. American Association of State Highway and Transport Officials.
BASt (1993). Schutzeinrichtungen an Bundesfernstrassen. Berichte der Bundesanstalt für Strassenwesen.Verkehrstechnik, Heft V6, Oktober 1993.
CEN/TC 226 N 185 E. Road restraint systems. Part 1: Terminology and general criteria for test methods.
Commission of the European Communities (1997). Promoting road safety in the EU; the Programme for1997-2001.
Czech Republic (1994). Testing and approval of safety barriers. Technical Specifications - TP60. Ministry ofTransport of the Czech Republic, January 1, 1994.
Department of Transport (1985). Safety fences and barriers. Departmental Standard 19/85. Department ofTransport, Highways and Traffic Directorate. United Kingdom
Durth (19987). Vergleich der Richtlinien für den Strassenentwurf in den Ländern der EuropäischenGemeinschaft. Forschungsauftrag: “Project Road Safety Year - Grant no VII-B-336” der Kommission derEuropäischen Gemeinschaft, Directorate General for Transport. Technische Hochschule Darmstadt, Oktober1987.
Elvik, R. (1994). The safety value of guardrails and crash cushions: a meta-analyses of evidence from evaluationstudies. Proceedings of the Conference Road Safety in Europe and Strategic Highway Research Programm,Lille, France, September 26-28, 1994.
ETSC (1997). Transport accident costs and the validation of life. European Transport Safety Council, Brussels.
Fer, B. (1993). Implantation complémentaire de glissières de sécurité en accotement. Autoroutes du sud de laFrance. Revue générale des routes et des aérodromes, No 704, février 1993.
Fer, B. (1995). Amélioration de la sécurité lors des sorties de chaussée en accotement. Autoroutes du sud de laFrance. Revue générale des routes et des aérodromes, février 1995.
FHA (1986). Roadside improvements for local roads and streets. Federal Highway Administration, Office ofHighway Safety.
Gülich, H.A. (1996). Stahl oder Beton im Mittelstreifen? Erfahrungen mit Schutzeinrichtungen sprechen fürfallbezogene Entscheidungen und angemessenen Einsatz von Beton. Strasse und Autobahn 12/1996.
Hall, J.W., Turner, D.S. & Hall, L.E. (1994). Concerns about use of severity indexes in roadside safetyevaluations. Transportation Research Record, No. 1468. Transportation Research Board, US.
INRETS (1993). Accidents sur autoroutes A6 - A7 - A9. Study LCB 9310, September 1993.
Kallberg, V-P (1994). Accident reducing potential of roadside improvements. Technical Research Centre ofFinland. Proceedings of the 22nd European Transport Forum, Seminar K Highways, 12 - 16 September 1994.PTRC.
115
Martin, J.L., Huet, R., Boissier, G., Bloch, P., Vergnes, I. & Laumon, B. (1997). The severity of primary impactwith metal or conrete central median barriers on French motorways. INRETS - LEAT. Preprint of theConference “Traffic Safety on Two Continents”, Lisbon, Portugal, 22-24 Septemer 1997.
Martin, J.L., Huet, R., Boissier, G., Bloch, P., Vergnes, I. & Laumon, B. (1997). Evualuation of theconsequences of systematically equipping the highway hard shoulder with safety barriers. 41st AnnualProceedings Association for the Advancement of Automotive Medicine, Orlando, Florida, November 10-11 1997.
McCarthy, L. (1987). Roadside safety. A national perspective. Public Roads, March 1987.
McDevitt, C.F. (1984). Recent innovations in traffic barriers and other roadside safety appurtenances.International Transport Congress, Sept. 23-27, 1984, Montreal.
McGee, H.W., Hughes, W.E. & Daily, K. (1995). Effect of highway standards on safety. NCHRP Report 374.National Cooperative Highway Research Program, Tranportation Research Board.
McLean, J. (1980). Review of rural road geometric standards. In: Proceedings of the Workshop on Economics ofRoad design Standards, Bureau of Transport Economics, Vol. 1, Canberra, Australia, 1980.
McNally, M.G. & Merheb, O. (1991). The impact of Jersey median barriers on the frequency of freewayaccidents. Institute of Transportation Studies, University of California. November, 1991.
M.E.T (1991). Caracteristiques routières et autoroutières. Circulaire No A/AW/205/91/02685. Ministère Wallonde l’Equipement et des Transports, M.E.T.
Michie, J.D. (1996). Roadside safety: areas of future focus. Transportation Research Circular, Number 453,1996.
O’Cinneide, D. (1994). A comparison of road design standards and operational regulations in Europe.University College, Cork, Ireland. Proceedings of the 22nd European Transport Forum, Seminar K Highways,12 - 16 September 1994. PTRC.
Ogden, W. (1997). The effects of paved shoulders on accidents on rural highways. Accident Analyses &Prevention, Vol. 29, No 3, 1997.
Pak-Poy and Kneebone Pty Ltd (1988). Road safety benefits from rural road improvements. Report CR71.Department of Transport and communications, Federal Office of Road Safety, Canberra, Australia, April 1988.
Pigman, J.G. & Agent, K.R. (1991). Guidelines for installation of guardrail. Transportation Research Record1302.
Pol, W.H.M. van de & Heijer, T. (1993). Optimalisatie van het profiel van een betonnen voertuigkering. R-93-14. SWOV Institute for Road Safety Research, Leidschendam. [in Dutch].
Pol, W.H.M. van de (1997). Literatuuronderzoek voertuigkering H4-niveau. R-97-49. SWOV Institute for RoadSafety Research, Leidschendam. [in Dutch].
Ramache, A. (1991). Roadside safety - A knowledge-base approach. Proceedings of the conference StrategicHighway Research and Traffic Safety on two continents, Gothenburg, Sweden, September 18 -20, 1991, Part 2.VTI-rapport, 372, Part 2.
116
Ray, M.H. (1994). Safety Advisor: Framework for performing roadside safety assessments. TransportationResearch Record, No. 1468. Transportation Research Board, US.
Research Results Digest (1997). Strategies for improving roadside safety. Research Results Digest, nr 220.Transportation Research Board, National Research Council, US, November 1997.
Ross, H.E. jr. (1995). Evolution roadside safety. Transportation Research Circular, Number 435, 1995.
Rijkswaterstaat/DHV (1990). Geleiderail of betonnen geleidebarrier in middenbermen van autosnelwegen.Rijkswaterstaat, Dienst Verkeerskunde, DHV Raadgevend Ingenieursbureau BV, maart 1990 [in Dutch].
Rijkswaterstaat (1998). Veilige inrichting van bermen. A draft of standards for motorways in the Netherlands.
RPS (1989). Richtlinien für passive Schutzeinrichtungen an Strassen. Forschungsgesellschaft für Strassen- undVerkehrswesen. Ausgabe 1989, Köln, Germany.
Ruyters, H.G.J.C.M., Slop, M. & Wegman, F.C.M. (Eds.). Safety effects of road design standards. R-94-7.SWOV Institute for Road Safety Research. Leidschendam, The Netherlands. 1994.
Schoon, C.C. (1985). Aanrijdingen met in stijfheid verschillende typen geleiderailconstructies; Een beschrijvingvan de ernst en mate van terugkaatsing van aanrijdingen tegen geleiderailconstructies. R-85-63. SWOVInstitute for Road Safety Research, Leidschendam. [in Dutch].
Schoon, C.C. (1990). After seven years RIMOB in practice; An evaluation of the Dutch impact attenuatorRIMOB. R-90-49. SWOV Institute for Road Safety Research, Leidschendam.
Schoon, C.C. & Bos, J.M.J. (1983). Boomongevallen; Een verkennend onderzoek naar de frequentie en ernstvan botsingen tegen obstakels in relatie tot de breedte van de obstakelvrije zone. R-83-23. SWOV Institute forRoad Safety Research, Leidschendam. [in Dutch].
Schoon, C.C. & Broertjes, P. (1995). Full scale test results of the RIMOB Crash Cushion; Description of testsand results conform standard CEN/TC 226/WG1. R-95-16. SWOV Institute for Road Safety Research,Leidschendam.
Schoon, C.C. & Edelman, A. (1978). Lighting columns; Research on the behaviour of lighting columns insideways-on and head-on impact tests with private cars. Publication 1978-2E. SWOV Institute for Road SafetyResearch, Leidschendam.
Schoon, C.C., Jordaan, D.J.R. & Pol, W.H.W. van de (1977). Praatpalen; Een nadere beschouwing van eenaantal oriënterende botsproeven die in opdracht van de Rijkswaterstaatswerkgroep 'Bermbeveiligingen' in 1971zijn gehouden op 'De Vlasakkers' te Amersfoort. SWOV, R-77-7.
Schoon, C.C. & Pol, W.H.W. van de (1987). Aflopende taluds; De invloed van diverse taludkenmerken op deafloop van taludincidenten, bepaald met behulp van mathematische simulaties. Deel 1: Gesimuleerdetaludincidenten zonder voertuigmanoeuvres. R-87-8. SWOV Institute for Road Safety Research, Leidschendam.[in Dutch].
Schoon, C.C. & Pol, W.H.M. van de (1988a). Aflopende taluds II; De invloed van diverse taludkenmerken opde afloop van taludincidenten, bepaald met behulp van mathematische simulaties; Deel II: Gesimuleerde talud-incidenten met voertuigmanoeuvres. R-88-15. SWOV Institute for Road Safety Research, Leidschendam. [inDutch].
Schoon, C.C. & Pol, W.H.M. van de (1988b). Opgaande taluds; De bepaling van acceptabele taludconfiguratiesop basis van de uitvoering van mathematische simulaties. R-88-27. SWOV Institute for Road Safety Research,Leidschendam. [in Dutch].
117
SETRA (1995). Highway design guide (except for motorways). Technical recommendations for the generaldesign and geometry of roads. Centre de la Sécurité et des Techniques Routières, France.
SETRA (1985). Instruction sur les conditions techniques d’amenagements des autoroutes de liaison. Centre dela Sécurité et des Techniques Routières, Direction des Routes, France.
Stoughton, (1996). An oldtimer suggests some activities for improving roadside safety. TransportationResearch Circular, Number 453, 1996.
SWOV (1985). De invloed van de wrijvingscoëfficiënt van betonnen geleideconstructies op de grootte van devoertuigvertraging en de klimhoogte van voorwielen. R-85-68. SWOV Institute for Road Safety Research,Leidschendam. [in Dutch].
Martens, M. H. and Kaptein N. A (1998). The effect of tunnel design on driving behaviour and traffic Safety: Aliterature review. SAFESTAR deliverable 2.1
Martens, M. H. and Kaptein N. A (1998).Guidelines and standards for tunnels on motorways. SAFESTARdeliverable 2.2
Martens, M. H. , Törnros J. and Kaptein N. A (1998). Effects of emergency lane, exits and entries and wallpatternd in tunnels on driving behaviour: Driving simulator studies. SAFESTAR deliverable 2.3.
Admunsen F.H. (1994). Studies of driver behaviour in Norwegian road tunnels. Tunnelling an undergroundspace technology, vol9, No. 1. Elsevier Science Ltd.
PIARC Committee on Road Tunnels. (1995). Road safety in tunnels.
PIARC, Technical committee on road tunnels (1983) XVII World Road Congress, Sydney, Australia 8-15October 1983.
PIARC, Technical committee on road tunnels (1987) XVIIIth World Road Congress, Brussels, Belgium 13-19September 1987.
6. Junctions
6.1 Express roads
Cardoso J. & Costa S. (1998) Comparative analysis of road safety on express road; report on workpackage 3.1 ofSAFESTAR. Laboratório Nacional de Engenharia Civil (LNEC), Lisbon (in preparation)
Ogden, K.W. (1996) Safer roads, a guide to road safeety engineering. Avebury Technical, Aldershot.
Hughes, W. & Amis, G. (1996) Accidents on rural roads: single carriageway ‘A’ class roads. Foundation forRoad Safety Research, Basingstoke.
Hughes, W., Amis, G. & Walford, A. (1996) Accidents on rural roads: dual carriageway ‘A’ class roads.Foundation for Road Safety Research, Basingstoke.
118
Slop, M., Jouineau, J.P., Machu, C., Weber, R., Bergh, T., Zaragoza Ramirez, A. & Janssen, T. (1996) Technical guide on road safety for interurban roads; INTERSAFE. European Road Safety Federation, Brussels.
ETSC (1995) Reducing traffic injuries resulting form excess and inappropriate speed. Euroepan TransportSafety Council, Brussels.
6.2 Major urban junctions
Brüde, U. & J. Larsson (1993). Model for predicting accidents at junctions where pedestrians and cyclists areinvolved. How well do they fit? In: Accident Analysis and Prevention. Volume 25. pp. 499-509.
Brüde, U. & J. Larsson (1996). The safety of cyclists at roundabouts. A comparison between Swedish, Danishand Dutch results. VTI meddelande. No. 810A. VTI Swedish National Road and Transport Research Institute,Linköping.
CROW (1997). Kruispunten buiten de bebouwde kom. Aanbevelingen voor toepassing middengeleiders.(Junctions outside built-up areas. Recommendations for the application of median strips). CROW, Ede.
Jadaan, K.S. & A.J. Nicholson (1992). Realtionships between road accidents and traffic flows in an urbannetwork. In: Traffic Engineering & Control. Volume 33. No. 9. pp. 507-511.
Kulmala, R. (1995). Safety at rural three- and four-arm junctions. Development and application of accidentprediction models. Ph.D. Thesis. Helsinki University of Technology & Technical Research Centre of FinlandVTT, Espoo.
Layfield, R.E.; I. Summersgill; R.D. Hall & K. Chatterjee (1996). Accidents at urban priority crossroads andstaggered junctions. TRL report 185. Transport Research Laboratory, Crowthorne.
Minnen, J. van (1995). Rotondes en voorrangsregelingen. (Roundabouts and priority regulations). R-95-58.Stichting Wetenschappelijk Onderzoek Verkeersveiligheid SWOV, Leidschendam.
Pickering, D.; R.D. Hall & M. Grimmer (1986). Accidents at rural T-junctions. Research Report 65.Tranportation and Road Research Laboratory, Crowthorne.
Road Directorate (1995). Uheldsmodeller for bygader. Del 1: Model for 3- og 4-benede kryds. Notat 22.Trafiksikkerhed og Miljø, Vejdirektoratet, Copenhagen.
Summersgill, I.; J.V. Kennedy & D. Baynes (1996). Accidents at three-arm priority junctions on urban single-carriageway roads. TRL report 184. Transport Research Laboratory, Crowthorne.
Tanner, J.C. (1953). Accidents at rural three-way junctions. In: Journal of Institution of Highway Engineers.July.
Taylor, M.C.; R.D. Hall & K. Chatterjee (1996). Accidents at 3-arm traffic signals on urban single-carriagewayroads. TRL report 135. Transport Research Laboratory, Crowthorne.
